notebooks/00_DGA_EDA.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "_L5b-Qddgw0s"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "from pprint import pprint\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from matplotlib import pyplot\n",
    "from unidecode import unidecode"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "mqSmS2V7D-k1"
   },
   "source": [
    "# Domain Generating Algorithms - EDA"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "oNFrWzLUN3I1"
   },
   "source": [
    "## Context:\n",
    "\n",
    "Once malicious software has been installed on a system it needs to be told what to do. To receive instructions the malware needs to communicate with a command and control server.\n",
    "Historically, the domain name of the command and control server was hardcoded into the code of the malware. This made it very easy for cybersecurity professionals to block this communication channel by blacklisting the domain once they discovered it.\n",
    "\n",
    "To avoid this weakness malware programs have started to use Domain Generating Algorithms (or DGAs).  These algorithms generate a number of random domain names where one of the generated domains is the control server. The infected computed scans through these domains trying to contact each, eventually it will try the correct domain. At this stage, the malware can receive instructions remotely.\n",
    "\n",
    "The dataset consists of domains that have been labeled:\n",
    "* `class` - is the domain `legit` or `dga`\n",
    "* `subclass` - where the `class` is not `legit` this will detail the type of malware that generated the domain. If `legit` the value for this will also be `legit`."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Uq-fAby8N0jd"
   },
   "source": [
    "## Objective\n",
    "\n",
    "If we assume that:\n",
    "* **True Postive**: Correctly identified a `dga` entry.\n",
    "* **True Negative**: Correctly identified a `legit` entry.\n",
    "* **False Positive**: Incorrectly identifed a `dga` entry as a `legit` entry - *Type 1 Error*.\n",
    "* **False Negative**: Incorrectly identified a `legit` entry as a `dga` entry  - *Type 2 Error*.\n",
    "\n",
    "We futher assume the **False Postive** as the worst outcome, that we allowed a program to establish a connection to a `dga`."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "EWgCy4Bc_x8Z"
   },
   "source": [
    "# Conclusions for Futher Work\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "LHGHZ_MRODMh"
   },
   "source": [
    "## Overall:\n",
    "\n",
    "* Accuracy alone is not a suffcient metric, rather F1 Score will give a better view of the performance in light of *Type 2 Errors* (False Negatives) being the most undesireable.\n",
    "* Vowel Counts do not provide a clear boundary between `class` entries. However *Vowel Ratios* have some discriminatory value between the `class` entries. This can be nominally improved if we consider only distinct characters.\n",
    "* There is some discriminating value in counting the number of digits, in that where the *Vowel Ratio* is zero `dga` domains have an increased likelyhood of having a zero digit count.\n",
    "* `domain` will be used as the input, as the additional information provided by `host` (e.g. Top Level Domain) would be difficult to action due to `class` imbalances."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "3F4VZ_dcN8-5"
   },
   "source": [
    "## Column Specific\n",
    "\n",
    "* `domain`\n",
    "    * Does not contain unique values, due to variation of `host` (e.g. companies with country specific sites), however the `class` and `subclass` values are the same across the duplicated observations of `domain`.\n",
    "    * The mimimum length character length of a `dga` entry is 12 characters. Furthermore `dga` entries have a greater variation in character length (`~5.94`) compared to `legit` (`~3.3`) - *this could be a useful feature*.\n",
    "* `host`\n",
    "    * Although it is possible to split out the Top Level Domain - `tld` (e.g. .com,.ru) the distribution is heavily skewed. Furthermore ~99% of the `tld` do not have observations where the `class` is `dga` - *thus as a feature `tld` would not generalise well*."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "3ScXPiOON4qi"
   },
   "source": [
    "## Considerations for Processing Steps\n",
    "\n",
    "* Convert to lowercase.\n",
    "* Normalise Unicode to ASCII characters sets.\n",
    "* Remove all characters that do not fall into the range of `[a-z0-9\\-]`.\n",
    "* Remove all whitespace.\n",
    "* Remove duplicates.\n",
    "* create `domain_len`, counting the number of characters in domain.\n",
    "* create `domain_vowel_count` and `domain_vowel_ratio` - `domain_vowel_count` is not for use the model but rather required for calculating the ratio of characters in the `domain` that are vowels.\n",
    "* create `domain_digit_count`, counting the number of characters in `domain` that are digits."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "QNButPHtObSN"
   },
   "source": [
    "# References\n",
    "\n",
    "![Example of URL parts](https://domain.me/wp-content/uploads/2016/10/IP-domain-1.jpg)\n",
    "\n",
    "Image Credit: [domain.me](https://domain.me/what-are-domain-names-how-do-they-work/) I will use the above definitions for referring constituiant parts of a \"domain\".\n",
    "\n",
    "For the canonical definition of URLs I referred to: [Domain names Implementation and Specification ietf.org](https://tools.ietf.org/html/rfc1035)\n",
    "\n",
    "For clarification on Unicode characters in DNS: [Internationalisation of DNS - unicode.org](https://unicode.org/faq/idn.html)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "AtEq-HGmi8r7"
   },
   "source": [
    "# Data Preview and Data Types"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "os.chdir(\"../\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "AMvWAET9gzI2"
   },
   "outputs": [],
   "source": [
    "data_df = pd.read_csv(\"data/raw/dga_domains.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "GgHSwF20i86i",
    "outputId": "7eb51dab-c70a-4fbd-b9f9-0a3c6b5975e4"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>host</th>\n",
       "      <th>domain</th>\n",
       "      <th>class</th>\n",
       "      <th>subclass</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>79135</th>\n",
       "      <td>myscoresoft.info</td>\n",
       "      <td>myscoresoft</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>60750</th>\n",
       "      <td>jdiqrjrrjvoitck.ru</td>\n",
       "      <td>jdiqrjrrjvoitck</td>\n",
       "      <td>dga</td>\n",
       "      <td>cryptolocker</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>121537</th>\n",
       "      <td>viet4rum.com</td>\n",
       "      <td>viet4rum</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21615</th>\n",
       "      <td>buzzstream.tv</td>\n",
       "      <td>buzzstream</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>87732</th>\n",
       "      <td>pbguide.com</td>\n",
       "      <td>pbguide</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                      host           domain  class      subclass\n",
       "79135     myscoresoft.info      myscoresoft  legit         legit\n",
       "60750   jdiqrjrrjvoitck.ru  jdiqrjrrjvoitck    dga  cryptolocker\n",
       "121537        viet4rum.com         viet4rum  legit         legit\n",
       "21615        buzzstream.tv       buzzstream  legit         legit\n",
       "87732          pbguide.com          pbguide  legit         legit"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df.sample(5, random_state = 9000)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "mBM5w8jgL3XL"
   },
   "source": [
    "# Observations on the Data Preview\n",
    "\n",
    "- `host` contains the SLD and the TLD, it does not seem to contain the subdomain e.g. \"www\".\n",
    "- `domain` is just the SLD.\n",
    "- `class` details if the record is legitimate (`legit`) or the result of a Domain Generating Algorithms (`dga`).\n",
    "- `subclass` where `class` is `dga` this column details the malware which used the domain. If the `class` is `legit` the value for the column is `legit`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "8-AgHdcpg-Wx",
    "outputId": "1c143bdb-924f-475b-9fd9-b4845efcc3b4"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(133926, 4)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 102
    },
    "colab_type": "code",
    "id": "P_nKuOSOg4Wr",
    "outputId": "8a5fa40b-01d8-403e-8a07-6fae4440ddd4"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "host        object\n",
       "domain      object\n",
       "class       object\n",
       "subclass    object\n",
       "dtype: object"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 102
    },
    "colab_type": "code",
    "id": "w2yYaXcbhyt2",
    "outputId": "24494f81-1d51-4981-9d14-e9ee05b8a28d"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "host        133926\n",
       "domain      133926\n",
       "class       133926\n",
       "subclass    133926\n",
       "dtype: int64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df.count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "urXS69lhFUuo"
   },
   "source": [
    "# Missing Values\n",
    "\n",
    "Before missing data checks, given we're working with only text columns lets eliminate leading and trailing whitespace, and then replace empty strings with a known value (`np.nan`)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "AeRgmmqwhdma"
   },
   "outputs": [],
   "source": [
    "for col in data_df.columns:\n",
    "    data_df[col] = data_df[col].apply(lambda x: str(x).strip())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "wS95qkQ8hWDq"
   },
   "outputs": [],
   "source": [
    "data_df = data_df.replace({ \"\": np.nan })"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 102
    },
    "colab_type": "code",
    "id": "g6qHJTfPg7n7",
    "outputId": "4ab0a0e8-5613-448a-9681-f6ae5b6f1593"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "host        133926\n",
       "domain      133926\n",
       "class       133926\n",
       "subclass    133926\n",
       "dtype: int64"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "cPA_oH6eN8uC"
   },
   "outputs": [],
   "source": [
    "assert(data_df.dropna(axis = 0, how = \"any\").shape == data_df.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "T63kpsM9N2mT"
   },
   "source": [
    "# Observations on Missing Values\n",
    "\n",
    "There appear to be no missing values in the dataset.\n",
    "\n",
    "Although we should check for empty inputs."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "MZbkmckMGT7u"
   },
   "source": [
    "# Uniqueness\n",
    "\n",
    "Based on my understanding of the dataset, I expect the `host` and the `domain` columns to contain unique values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 85
    },
    "colab_type": "code",
    "id": "EfMXpcpHF-5x",
    "outputId": "9c8837f7-d395-4707-a9ed-0530c0621f69"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "host is unique: True\n",
      "domain is unique: False\n",
      "class is unique: False\n",
      "subclass is unique: False\n"
     ]
    }
   ],
   "source": [
    "for col in data_df.columns:\n",
    "    print(col, \"is unique:\", data_df[col].is_unique)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "YDWQgu34G31p"
   },
   "source": [
    "It appears I was mistaken, lets see what is in domain."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 359
    },
    "colab_type": "code",
    "id": "IIwBRwu7Gmd5",
    "outputId": "344ace6f-63bc-43c0-bd04-0c42f78f7e15"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>domain</th>\n",
       "      <th>occurances</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>provincia</td>\n",
       "      <td>47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>travian</td>\n",
       "      <td>46</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>groupon</td>\n",
       "      <td>46</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>trivago</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>philips</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>careerjet</td>\n",
       "      <td>45</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>mcdonalds</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>mercedes-benz</td>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>tripadvisor</td>\n",
       "      <td>40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>skyscanner</td>\n",
       "      <td>37</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          domain  occurances\n",
       "0      provincia          47\n",
       "1        travian          46\n",
       "2        groupon          46\n",
       "3        trivago          45\n",
       "4        philips          45\n",
       "5      careerjet          45\n",
       "6      mcdonalds          41\n",
       "7  mercedes-benz          40\n",
       "8    tripadvisor          40\n",
       "9     skyscanner          37"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domain_values_counts_df = pd.DataFrame(data_df[\"domain\"].value_counts().sort_values(ascending = False)).reset_index()\n",
    "domain_values_counts_df.columns = [\"domain\", \"occurances\"]\n",
    "domain_values_counts_df.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "vuMo2-ddHYjr"
   },
   "source": [
    "What is the distribution of the duplication of `domain`?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 170
    },
    "colab_type": "code",
    "id": "GrwY5_-gHMKV",
    "outputId": "efd6125d-7e69-4c90-86ee-964ce022da7e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    114525.000000\n",
       "mean          1.169404\n",
       "std           1.076192\n",
       "min           1.000000\n",
       "25%           1.000000\n",
       "50%           1.000000\n",
       "75%           1.000000\n",
       "max          47.000000\n",
       "Name: occurances, dtype: float64"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domain_values_counts_df[\"occurances\"].describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "_viWTnwnG-96"
   },
   "source": [
    "It would seem for the majority of the time host and domain are similar. Let's peek at the first 10 entries in our dataset with the most frequently occuring `domain` value."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 359
    },
    "colab_type": "code",
    "id": "RnvkbEvCHnQ9",
    "outputId": "3d82975f-d4e2-4b61-8e3d-cec35f403dda"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>host</th>\n",
       "      <th>domain</th>\n",
       "      <th>class</th>\n",
       "      <th>subclass</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>92620</th>\n",
       "      <td>provincia.genova.it</td>\n",
       "      <td>provincia</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92621</th>\n",
       "      <td>provincia.teramo.it</td>\n",
       "      <td>provincia</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92622</th>\n",
       "      <td>provincia.cs.it</td>\n",
       "      <td>provincia</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92623</th>\n",
       "      <td>provincia.so.it</td>\n",
       "      <td>provincia</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92624</th>\n",
       "      <td>provincia.fe.it</td>\n",
       "      <td>provincia</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92625</th>\n",
       "      <td>provincia.udine.it</td>\n",
       "      <td>provincia</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92626</th>\n",
       "      <td>provincia.lecco.it</td>\n",
       "      <td>provincia</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92627</th>\n",
       "      <td>provincia.le.it</td>\n",
       "      <td>provincia</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92628</th>\n",
       "      <td>provincia.treviso.it</td>\n",
       "      <td>provincia</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92629</th>\n",
       "      <td>provincia.tn.it</td>\n",
       "      <td>provincia</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       host     domain  class subclass\n",
       "92620   provincia.genova.it  provincia  legit    legit\n",
       "92621   provincia.teramo.it  provincia  legit    legit\n",
       "92622       provincia.cs.it  provincia  legit    legit\n",
       "92623       provincia.so.it  provincia  legit    legit\n",
       "92624       provincia.fe.it  provincia  legit    legit\n",
       "92625    provincia.udine.it  provincia  legit    legit\n",
       "92626    provincia.lecco.it  provincia  legit    legit\n",
       "92627       provincia.le.it  provincia  legit    legit\n",
       "92628  provincia.treviso.it  provincia  legit    legit\n",
       "92629       provincia.tn.it  provincia  legit    legit"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[data_df[\"domain\"] == domain_values_counts_df.iloc[0][\"domain\"]].head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "I0RqdF1mInzo"
   },
   "source": [
    "Based on the above it would seem that the `host` column contains both the Second Level Domain (SLD) and the Top Level Domain (TLD).\n",
    "\n",
    "From this there are a couple of assumptions:\n",
    "* Splitting the `host` field on `.` characters and taking the first slice will result in a value equal to the domain.\n",
    "* For the many entries that share the same domain, they will belong to the same `class` and `subclass`.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "KrnAfGNSInVP"
   },
   "outputs": [],
   "source": [
    "data_sld = data_df[\"host\"].apply(lambda x: x.split(\".\")[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ryNJb2WaKoAW"
   },
   "source": [
    "Let's validate the first assumption."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "0sHU1ldcKeG6"
   },
   "outputs": [],
   "source": [
    "assert(np.all(data_sld == data_df[\"domain\"]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "VOhvGXemKrTf"
   },
   "source": [
    "Now for the second. That for grouped by `host` the number of unique values of `class` and `subclass` will equal one."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "NRiBkB1IKtan"
   },
   "outputs": [],
   "source": [
    "data_grouped_by_domain_sum_df = data_df[[\"domain\", \"class\", \"subclass\"]].groupby(\"domain\").agg(pd.Series.nunique)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 235
    },
    "colab_type": "code",
    "id": "cMNPrCRULFm4",
    "outputId": "cd3f553c-b05e-4094-be0e-5a8538025eee"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>class</th>\n",
       "      <th>subclass</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>000directory</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>000webhost</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>001fans</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>01-telecharger</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>010shangpu</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                class  subclass\n",
       "domain                         \n",
       "000directory        1         1\n",
       "000webhost          1         1\n",
       "001fans             1         1\n",
       "01-telecharger      1         1\n",
       "010shangpu          1         1"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_grouped_by_domain_sum_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 450
    },
    "colab_type": "code",
    "id": "fvy9DZ2XTM8P",
    "outputId": "b5f49beb-6638-4f20-e02c-d11b3ba1d56a"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>class</th>\n",
       "      <th>subclass</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>000directory</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pingwest</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pinkems</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pinkemma</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pinkdino</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>freshupnow</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>freshtv</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>freshthemes</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>freshteenvideos</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>zzztube</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>114525 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                 class  subclass\n",
       "domain                          \n",
       "000directory         1         1\n",
       "pingwest             1         1\n",
       "pinkems              1         1\n",
       "pinkemma             1         1\n",
       "pinkdino             1         1\n",
       "...                ...       ...\n",
       "freshupnow           1         1\n",
       "freshtv              1         1\n",
       "freshthemes          1         1\n",
       "freshteenvideos      1         1\n",
       "zzztube              1         1\n",
       "\n",
       "[114525 rows x 2 columns]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_grouped_by_domain_sum_df.sort_values(\"class\", ascending = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "VVbIFitCLXxQ"
   },
   "outputs": [],
   "source": [
    "assert(data_grouped_by_domain_sum_df[\"class\"].max() == 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "DdGYxPasLdTy"
   },
   "outputs": [],
   "source": [
    "assert(data_grouped_by_domain_sum_df[\"subclass\"].max() == 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "e65TatruPCcD"
   },
   "source": [
    "# Observations on Uniqueness\n",
    "\n",
    "* `domain` is **not unique**, if it is considered as part of model it would be prudent to remove the duplicated values."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "awr-k57rK_Ut"
   },
   "source": [
    "# Data Quality Checks\n",
    "\n",
    "Given we've got mostly string data let's review the assumptions about the data pertaining to the quality of the data:\n",
    "\n",
    "* No control characters, such at linebreak or tabs should be present.\n",
    "* The characterset will be ascii.\n",
    "* There should be a variety of characters \n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "nZwTUhrjWQul"
   },
   "outputs": [],
   "source": [
    "validate_text_columns = [\"host\", \"domain\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "YLK9WIlDUysZ"
   },
   "source": [
    "## No Control Characters\n",
    "\n",
    "If we assume that the range of expected characters to be `a-z`, `0-9`, `-`, then we can compare the ordinal values in our dataset to check if they are in this range."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "T0NWpg4NKBX_"
   },
   "source": [
    "Let's first determine the numeric ranges we need for ascii characters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 136
    },
    "colab_type": "code",
    "id": "dD8_uZXzJAFa",
    "outputId": "696c119f-9b77-4670-af2f-1630047020b9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "- 45\n",
      ". 46\n",
      "0 48\n",
      "1 49\n",
      "9 57\n",
      "a 97\n",
      "z 122\n"
     ]
    }
   ],
   "source": [
    "ascii_ranges = [\"-\", \".\", \"0\", \"1\", \"9\", \"a\", \"z\"]\n",
    "ascii_ord_ranges = list()\n",
    "for c in ascii_ranges:\n",
    "    ascii_ord_ranges.append(ord(c))\n",
    "    print(c, ord(c))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "rp6U6rX_KAiX"
   },
   "source": [
    "Lets create our list of `expected_ords` and test the characters we identified for `ascii_ranges` have their numeric representation therein."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "_XXWTI3jJPpj"
   },
   "outputs": [],
   "source": [
    "expected_ords = [ord(\"-\"), ord(\".\")] + list(range(ord(\"0\"), ord(\"9\")+1)) + list(range(ord(\"a\"), ord(\"z\")+ 1))\n",
    "\n",
    "# a simple check to ensure the range domains we identified are present\n",
    "for i in ascii_ord_ranges:\n",
    "    assert(i in expected_ords)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "19m-isGQKVuo"
   },
   "source": [
    "Now to create a new dataframe containing text in domain represented as a sequence of number representing the characters.\n",
    "\n",
    "Also taking the additional step of converting the text to lower, this is step that will need to be taken in preprocessing."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 445
    },
    "colab_type": "code",
    "id": "Yk8apIfLGPc0",
    "outputId": "4f1cb470-d9fe-42c1-87eb-3fc5c7369174"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "host\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>50</th>\n",
       "      <th>51</th>\n",
       "      <th>52</th>\n",
       "      <th>53</th>\n",
       "      <th>54</th>\n",
       "      <th>55</th>\n",
       "      <th>56</th>\n",
       "      <th>57</th>\n",
       "      <th>58</th>\n",
       "      <th>59</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>48.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>105.0</td>\n",
       "      <td>114.0</td>\n",
       "      <td>101.0</td>\n",
       "      <td>99.0</td>\n",
       "      <td>116.0</td>\n",
       "      <td>111.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>48.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>119.0</td>\n",
       "      <td>101.0</td>\n",
       "      <td>98.0</td>\n",
       "      <td>104.0</td>\n",
       "      <td>111.0</td>\n",
       "      <td>115.0</td>\n",
       "      <td>116.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>48.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>102.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>110.0</td>\n",
       "      <td>115.0</td>\n",
       "      <td>46.0</td>\n",
       "      <td>99.0</td>\n",
       "      <td>111.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>48.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>45.0</td>\n",
       "      <td>116.0</td>\n",
       "      <td>101.0</td>\n",
       "      <td>108.0</td>\n",
       "      <td>101.0</td>\n",
       "      <td>99.0</td>\n",
       "      <td>104.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>48.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>115.0</td>\n",
       "      <td>104.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>110.0</td>\n",
       "      <td>103.0</td>\n",
       "      <td>112.0</td>\n",
       "      <td>117.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 60 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     0     1     2      3      4      5      6      7      8      9   ...  50  \\\n",
       "0  48.0  48.0  48.0  100.0  105.0  114.0  101.0   99.0  116.0  111.0  ... NaN   \n",
       "1  48.0  48.0  48.0  119.0  101.0   98.0  104.0  111.0  115.0  116.0  ... NaN   \n",
       "2  48.0  48.0  49.0  102.0   97.0  110.0  115.0   46.0   99.0  111.0  ... NaN   \n",
       "3  48.0  49.0  45.0  116.0  101.0  108.0  101.0   99.0  104.0   97.0  ... NaN   \n",
       "4  48.0  49.0  48.0  115.0  104.0   97.0  110.0  103.0  112.0  117.0  ... NaN   \n",
       "\n",
       "   51  52  53  54  55  56  57  58  59  \n",
       "0 NaN NaN NaN NaN NaN NaN NaN NaN NaN  \n",
       "1 NaN NaN NaN NaN NaN NaN NaN NaN NaN  \n",
       "2 NaN NaN NaN NaN NaN NaN NaN NaN NaN  \n",
       "3 NaN NaN NaN NaN NaN NaN NaN NaN NaN  \n",
       "4 NaN NaN NaN NaN NaN NaN NaN NaN NaN  \n",
       "\n",
       "[5 rows x 60 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "domain\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>46</th>\n",
       "      <th>47</th>\n",
       "      <th>48</th>\n",
       "      <th>49</th>\n",
       "      <th>50</th>\n",
       "      <th>51</th>\n",
       "      <th>52</th>\n",
       "      <th>53</th>\n",
       "      <th>54</th>\n",
       "      <th>55</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>48.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>105.0</td>\n",
       "      <td>114.0</td>\n",
       "      <td>101.0</td>\n",
       "      <td>99.0</td>\n",
       "      <td>116.0</td>\n",
       "      <td>111.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>48.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>119.0</td>\n",
       "      <td>101.0</td>\n",
       "      <td>98.0</td>\n",
       "      <td>104.0</td>\n",
       "      <td>111.0</td>\n",
       "      <td>115.0</td>\n",
       "      <td>116.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>48.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>102.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>110.0</td>\n",
       "      <td>115.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>48.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>45.0</td>\n",
       "      <td>116.0</td>\n",
       "      <td>101.0</td>\n",
       "      <td>108.0</td>\n",
       "      <td>101.0</td>\n",
       "      <td>99.0</td>\n",
       "      <td>104.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>48.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>48.0</td>\n",
       "      <td>115.0</td>\n",
       "      <td>104.0</td>\n",
       "      <td>97.0</td>\n",
       "      <td>110.0</td>\n",
       "      <td>103.0</td>\n",
       "      <td>112.0</td>\n",
       "      <td>117.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 56 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     0     1     2      3      4      5      6      7      8      9   ...  46  \\\n",
       "0  48.0  48.0  48.0  100.0  105.0  114.0  101.0   99.0  116.0  111.0  ... NaN   \n",
       "1  48.0  48.0  48.0  119.0  101.0   98.0  104.0  111.0  115.0  116.0  ... NaN   \n",
       "2  48.0  48.0  49.0  102.0   97.0  110.0  115.0    NaN    NaN    NaN  ... NaN   \n",
       "3  48.0  49.0  45.0  116.0  101.0  108.0  101.0   99.0  104.0   97.0  ... NaN   \n",
       "4  48.0  49.0  48.0  115.0  104.0   97.0  110.0  103.0  112.0  117.0  ... NaN   \n",
       "\n",
       "   47  48  49  50  51  52  53  54  55  \n",
       "0 NaN NaN NaN NaN NaN NaN NaN NaN NaN  \n",
       "1 NaN NaN NaN NaN NaN NaN NaN NaN NaN  \n",
       "2 NaN NaN NaN NaN NaN NaN NaN NaN NaN  \n",
       "3 NaN NaN NaN NaN NaN NaN NaN NaN NaN  \n",
       "4 NaN NaN NaN NaN NaN NaN NaN NaN NaN  \n",
       "\n",
       "[5 rows x 56 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ord_data_dfs = dict()\n",
    "\n",
    "for column in validate_text_columns:\n",
    "    print(column)\n",
    "    ord_data_dfs[column] = data_df[column].apply(lambda x: [ord(w) for w in x.lower()]).apply(pd.Series)\n",
    "    display(ord_data_dfs[column].head())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "dRGsG6wzLGGC"
   },
   "source": [
    "Let's replace the nothing values with a known number that we will permit to be present."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "vp-C1sMwLMaT"
   },
   "outputs": [],
   "source": [
    "for column in validate_text_columns:\n",
    "      ord_data_dfs[column] = ord_data_dfs[column].fillna(-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 119
    },
    "colab_type": "code",
    "id": "Ey_PvAVlKtis",
    "outputId": "411e02fe-94e6-4089-be3b-186fa16a252e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "host\n",
      "host - column: 2 contained unknown ordinal value\n",
      "\n",
      "domain\n",
      "domain - column: 2 contained unknown ordinal value\n",
      "\n"
     ]
    }
   ],
   "source": [
    "unknown_ord_columns = dict()\n",
    "\n",
    "for column in validate_text_columns:\n",
    "    print(column)\n",
    "    unknown_ord_columns[column] = set()\n",
    "\n",
    "    for ord_column in ord_data_dfs[column].columns:\n",
    "        try:\n",
    "            assert(all(ord_data_dfs[column][ord_column].isin(expected_ords + [-1])))\n",
    "        except AssertionError:\n",
    "            print(column, \"- column:\", ord_column, \"contained unknown ordinal value\")\n",
    "            unknown_ord_columns[column].add(ord_column)\n",
    "            continue\n",
    "      \n",
    "    print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "K2VL5DqHLo_O"
   },
   "source": [
    "What did we miss?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 221
    },
    "colab_type": "code",
    "id": "_OefeEmXbHbt",
    "outputId": "512929c5-80ec-4a2a-8de6-661a479aefcf"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "host\n",
      "99692    10.0\n",
      "Name: 2, dtype: float64\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "99692    ru\\ndercxjvpfswhmtwtcfycaijdufa.ru\n",
       "Name: host, dtype: object"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "domain\n",
      "99692    10.0\n",
      "Name: 2, dtype: float64\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "99692    ru\\ndercxjvpfswhmtwtcfycaijdufa\n",
       "Name: domain, dtype: object"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "for column in validate_text_columns:\n",
    "    print(column)\n",
    "    for ord_column in unknown_ord_columns[column]:    \n",
    "        unknown_ord_index = (~ord_data_dfs[column][ord_column].isin(expected_ords + [-1]))\n",
    "        if any(unknown_ord_index):\n",
    "            print(ord_data_dfs[column].loc[unknown_ord_index][ord_column])\n",
    "            display(data_df.loc[unknown_ord_index][column])\n",
    "\n",
    "    print()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ordinal `10` is a newline."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "AUcE1VGlLrCr",
    "outputId": "7c537ab5-3416-48aa-ae71-3350a5302081"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\n'"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chr(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "s8-w4rRkL9xP"
   },
   "source": [
    "## Observations on Control Characters \n",
    "It would seem we have a line break in our dataset. This could be the result of malformed encoding, or some data capture error.\n",
    "\n",
    "For preprocessing we should: \n",
    "* Convert `domain` and `host` to lowercase\n",
    "* Remove all characters that do not fall within the expected ordinal values for domain.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "mV2GhSr7RFBp"
   },
   "source": [
    "## Non-ASCII characters\n",
    "\n",
    "Although we've identified a range of characters we will permit, it does assume that we're operating in an ASCII characters set. If we remove all characters that don't meet this criteria we'll have a fair bit of potential information loss for example: `é` could be viewed as `e`.\n",
    "\n",
    "Unicode support has been implemented in dns names as per https://unicode.org/faq/idn.html\n",
    "However for the purposes of dimension reduction we'll assume an ASCII character-set.\n",
    "\n",
    "\n",
    "Create a normalised copy of the two columns of interest `domain` and `host`, and then compare them to the exisiting columns.\n",
    "\n",
    "If using Python 3.7+ one could use: https://docs.python.org/3/library/stdtypes.html#str.isascii"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "iwkSnGQCQCLe",
    "outputId": "fdc57e89-8aea-4557-da75-533d58f3f79b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Python 3.7.1\n"
     ]
    }
   ],
   "source": [
    "! python --version"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "f9LawlgzNO_I"
   },
   "outputs": [],
   "source": [
    "unidecode_data = dict()\n",
    "\n",
    "for column in validate_text_columns:\n",
    "    unidecode_data[column] = data_df[column].apply(unidecode)\n",
    "    try:\n",
    "        assert(all(unidecode_data[column] == data_df[column]))\n",
    "    except AssertionError as e: \n",
    "        print(column, \"contains non-ascii characters\")\n",
    "        raise e"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "FiusjtM8qGCF"
   },
   "source": [
    "## Observations on ASCII Characters\n",
    "\n",
    "While the provided data appears to be within the ASCII character set, we cannot assume this for further inputs.\n",
    "\n",
    "Normalisation to ASCII will be part of future inputs."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "UYMDndY6h-Gp"
   },
   "source": [
    "# Distributions of Categorical Data\n",
    "\n",
    "Let's review the occurances of `class` in the dataset."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "rqeWIefqRhvl"
   },
   "source": [
    "## `class`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 68
    },
    "colab_type": "code",
    "id": "JrwlDyJRhSrs",
    "outputId": "a776f61a-8cc2-4d2d-8653-490cf4d7c98c"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "legit    81261\n",
       "dga      52665\n",
       "Name: class, dtype: int64"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[\"class\"].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "363OCMxfQcX9"
   },
   "source": [
    "As a ratio?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 111
    },
    "colab_type": "code",
    "id": "vkoRenvNiZmn",
    "outputId": "b05f1619-25a2-49da-88cb-f343e29a1d15"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>0.60676</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dga</th>\n",
       "      <td>0.39324</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         class\n",
       "legit  0.60676\n",
       "dga    0.39324"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame(data_df[\"class\"].value_counts()) / len(data_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 294
    },
    "colab_type": "code",
    "id": "jJ3BElq8iNm-",
    "outputId": "d4ccf7f2-673e-458d-8a19-d7e996ad26a4"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x209576c87b8>"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_df[\"class\"].value_counts().plot.bar()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "pTqxQi7zRqdx"
   },
   "source": [
    "## Observations on `class` Catagorical Data\n",
    "While not perfectly balanced, there isn't a great imbalance between the two classes. Thus using stratified sampling should be a viable sampling method, without misrepresenting the ground truth.\n",
    "\n",
    "For further robustness additional scoring metrics apart from accuracy should be used, as if our prior assumption around type 1 error (False Positive) being a worse outcome than a type 2 error, accuracy is not sufficient. Given that we have an imbalanced dataset we should consider another metric that accounts for this such as `f1-score`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "iFb5g_OmRVlP"
   },
   "source": [
    "## `subclass`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 102
    },
    "colab_type": "code",
    "id": "b3B7whCdiD1_",
    "outputId": "dffa5023-71fe-4899-da6b-0667f241500f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "legit           81261\n",
       "cryptolocker    34319\n",
       "newgoz          10999\n",
       "goz              7347\n",
       "Name: subclass, dtype: int64"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[\"subclass\"].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 173
    },
    "colab_type": "code",
    "id": "iYZauo3NiheR",
    "outputId": "843349e1-3a8a-4907-934e-6a2e1b0a48a7"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>subclass</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>0.606760</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cryptolocker</th>\n",
       "      <td>0.256253</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>newgoz</th>\n",
       "      <td>0.082127</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>goz</th>\n",
       "      <td>0.054859</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              subclass\n",
       "legit         0.606760\n",
       "cryptolocker  0.256253\n",
       "newgoz        0.082127\n",
       "goz           0.054859"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame(data_df[\"subclass\"].value_counts()) / len(data_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 334
    },
    "colab_type": "code",
    "id": "0bxRsCq-iHOt",
    "outputId": "8986ede9-83aa-4cbd-f7a5-063bac61d958"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x209659efd30>"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAEsCAYAAADD8sRQAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAc/klEQVR4nO3df7ic5V3n8fenSX/QViiBwNIEm1hiFdDSkkW0/myspD9scBc01ZasRtMi2ta6q8HLvbrqFRd218XFXbCpWAJqIdB2SW1ppcH+cMXgoaVNA+XiCBSyieQUKMUq1OB3/5j7bCeH4Zw5JyeZOYf367rmmpnvPPfk+8zV8jnP/cwzd6oKSZKeNegGJEnDwUCQJAEGgiSpMRAkSYCBIElqFg66gZk69thja9myZYNuQ5LmlNtuu+0rVbW412tzNhCWLVvGyMjIoNuQpDklyZef7jWnjCRJgIEgSWoMBEkSYCBIkhoDQZIEGAiSpMZAkCQBBoIkqTEQJElAn1cqJ/kV4OeBAnYCPws8H7gWWAbcB/xkVT3Str8QWA88Cby9qj7e6qcDVwJHAB8F3lFVleS5wFXA6cBDwE9V1X2zsYPTsWzjRw73Pzkj9130+kG3IGkemvIIIckS4O3Ayqo6FVgArAU2AturagWwvT0nycnt9VOA1cBlSRa0t7sc2ACsaLfVrb4eeKSqTgIuAS6elb2TJPWt3ymjhcARSRbSOTLYA6wBtrTXtwBnt8drgGuq6omquhcYBc5IcgJwZFXdUp11O6+aMGb8va4HViXJQeyXJGmapgyEqvq/wH8D7gf2Ao9W1V8Ax1fV3rbNXuC4NmQJ8EDXW+xutSXt8cT6AWOqaj/wKHDMxF6SbEgykmRkbGys332UJPWhnymjo+n8Bb8ceDHwgiRvnmxIj1pNUp9szIGFqs1VtbKqVi5e3PPXWyVJM9TPlNGPAvdW1VhV/TPwQeD7gAfbNBDtfl/bfjdwYtf4pXSmmHa3xxPrB4xp01JHAQ/PZIckSTPTTyDcD5yZ5PltXn8VcCewDVjXtlkH3NAebwPWJnlukuV0Th7f2qaVHktyZnuf8yaMGX+vc4Cb23kGSdJhMuXXTqtqR5Lrgc8C+4HPAZuBFwJbk6ynExrntu13JdkK3NG2v6Cqnmxvdz7f/Nrpje0GcAVwdZJROkcGa2dl7yRJfevrOoSqejfw7gnlJ+gcLfTafhOwqUd9BDi1R/1xWqBIkgbDK5UlSYCBIElqDARJEmAgSJIaA0GSBBgIkqTGQJAkAQaCJKkxECRJgIEgSWoMBEkSYCBIkhoDQZIEGAiSpMZAkCQBBoIkqZkyEJK8LMntXbevJXlnkkVJbkpyd7s/umvMhUlGk9yV5Kyu+ulJdrbXLm1LadKW27y21XckWXYodlaS9PSmDISququqTquq04DTgX8EPgRsBLZX1Qpge3tOkpPpLIF5CrAauCzJgvZ2lwMb6KyzvKK9DrAeeKSqTgIuAS6end2TJPVrulNGq4C/q6ovA2uALa2+BTi7PV4DXFNVT1TVvcAocEaSE4Ajq+qWqirgqgljxt/remDV+NGDJOnwmG4grAXe3x4fX1V7Adr9ca2+BHiga8zuVlvSHk+sHzCmqvYDjwLHTPzHk2xIMpJkZGxsbJqtS5Im03cgJHkO8Ebguqk27VGrSeqTjTmwULW5qlZW1crFixdP0YYkaTqmc4TwWuCzVfVge/5gmwai3e9r9d3AiV3jlgJ7Wn1pj/oBY5IsBI4CHp5Gb5KkgzSdQHgT35wuAtgGrGuP1wE3dNXXtm8OLadz8vjWNq30WJIz2/mB8yaMGX+vc4Cb23kGSdJhsrCfjZI8H3gN8Nau8kXA1iTrgfuBcwGqaleSrcAdwH7ggqp6so05H7gSOAK4sd0ArgCuTjJK58hg7UHskyRpBvoKhKr6Ryac5K2qh+h866jX9puATT3qI8CpPeqP0wJFkjQYXqksSQIMBElSYyBIkgADQZLUGAiSJMBAkCQ1BoIkCTAQJEmNgSBJAgwESVJjIEiSAANBktQYCJIkwECQJDUGgiQJMBAkSY2BIEkC+gyEJC9Kcn2SLyW5M8n3JlmU5KYkd7f7o7u2vzDJaJK7kpzVVT89yc722qVtbWXa+svXtvqOJMtme0clSZPr9wjhfwAfq6rvAF4O3AlsBLZX1Qpge3tOkpPprIl8CrAauCzJgvY+lwMbgBXttrrV1wOPVNVJwCXAxQe5X5KkaZoyEJIcCfwgcAVAVX2jqr4KrAG2tM22AGe3x2uAa6rqiaq6FxgFzkhyAnBkVd1SVQVcNWHM+HtdD6waP3qQJB0e/RwhfBswBrwvyeeS/FGSFwDHV9VegHZ/XNt+CfBA1/jdrbakPZ5YP2BMVe0HHgWOmdhIkg1JRpKMjI2N9bmLkqR+9BMIC4FXApdX1SuAr9Omh55Gr7/sa5L6ZGMOLFRtrqqVVbVy8eLFk3ctSZqWfgJhN7C7qna059fTCYgH2zQQ7X5f1/Yndo1fCuxp9aU96geMSbIQOAp4eLo7I0mauSkDoar+HnggyctaaRVwB7ANWNdq64Ab2uNtwNr2zaHldE4e39qmlR5LcmY7P3DehDHj73UOcHM7zyBJOkwW9rndLwN/muQ5wD3Az9IJk61J1gP3A+cCVNWuJFvphMZ+4IKqerK9z/nAlcARwI3tBp0T1lcnGaVzZLD2IPdLkjRNfQVCVd0OrOzx0qqn2X4TsKlHfQQ4tUf9cVqgSJIGwyuVJUmAgSBJagwESRJgIEiSGgNBkgQYCJKkxkCQJAEGgiSpMRAkSYCBIElqDARJEmAgSJIaA0GSBBgIkqTGQJAkAQaCJKnpKxCS3JdkZ5Lbk4y02qIkNyW5u90f3bX9hUlGk9yV5Kyu+untfUaTXNqW0qQtt3ltq+9Ismx2d1OSNJXpHCH8SFWdVlXjK6dtBLZX1Qpge3tOkpPpLIF5CrAauCzJgjbmcmADnXWWV7TXAdYDj1TVScAlwMUz3yVJ0kwczJTRGmBLe7wFOLurfk1VPVFV9wKjwBlJTgCOrKpbqqqAqyaMGX+v64FV40cPkqTDo99AKOAvktyWZEOrHV9VewHa/XGtvgR4oGvs7lZb0h5PrB8wpqr2A48Cx0xsIsmGJCNJRsbGxvpsXZLUj4V9bveqqtqT5DjgpiRfmmTbXn/Z1yT1ycYcWKjaDGwGWLly5VNelyTNXF9HCFW1p93vAz4EnAE82KaBaPf72ua7gRO7hi8F9rT60h71A8YkWQgcBTw8/d2RJM3UlIGQ5AVJvmX8MfBjwBeBbcC6ttk64Ib2eBuwtn1zaDmdk8e3tmmlx5Kc2c4PnDdhzPh7nQPc3M4zSJIOk36mjI4HPtTO8S4E/qyqPpbkb4GtSdYD9wPnAlTVriRbgTuA/cAFVfVke6/zgSuBI4Ab2w3gCuDqJKN0jgzWzsK+SZKmYcpAqKp7gJf3qD8ErHqaMZuATT3qI8CpPeqP0wJFkjQYXqksSQIMBElSYyBIkgADQZLUGAiSJMBAkCQ1BoIkCTAQJEmNgSBJAgwESVJjIEiSAANBktQYCJIkwECQJDUGgiQJMBAkSY2BIEkCphEISRYk+VySP2/PFyW5Kcnd7f7orm0vTDKa5K4kZ3XVT0+ys712aVtbmbb+8rWtviPJstnbRUlSP6ZzhPAO4M6u5xuB7VW1AtjenpPkZDprIp8CrAYuS7Kgjbkc2ACsaLfVrb4eeKSqTgIuAS6e0d5Ikmasr0BIshR4PfBHXeU1wJb2eAtwdlf9mqp6oqruBUaBM5KcABxZVbdUVQFXTRgz/l7XA6vGjx4kSYdHv0cIvw/8GvAvXbXjq2ovQLs/rtWXAA90bbe71Za0xxPrB4ypqv3Ao8AxE5tIsiHJSJKRsbGxPluXJPVjykBI8gZgX1Xd1ud79vrLviapTzbmwELV5qpaWVUrFy9e3Gc7kqR+LOxjm1cBb0zyOuB5wJFJ/gR4MMkJVbW3TQfta9vvBk7sGr8U2NPqS3vUu8fsTrIQOAp4eIb7JEmagSmPEKrqwqpaWlXL6Jwsvrmq3gxsA9a1zdYBN7TH24C17ZtDy+mcPL61TSs9luTMdn7gvAljxt/rnPZvPOUIQZJ06PRzhPB0LgK2JlkP3A+cC1BVu5JsBe4A9gMXVNWTbcz5wJXAEcCN7QZwBXB1klE6RwZrD6IvSdIMTCsQquqTwCfb44eAVU+z3SZgU4/6CHBqj/rjtECRJA2GVypLkgADQZLUGAiSJMBAkCQ1BoIkCTAQJEmNgSBJAgwESVJzMFcqS5NatvEjg26hL/dd9PpBtyANBY8QJEmAgSBJagwESRJgIEiSGgNBkgQYCJKkxkCQJAF9BEKS5yW5Ncnnk+xK8lutvijJTUnubvdHd425MMlokruSnNVVPz3JzvbapW0pTdpym9e2+o4ky2Z/VyVJk+nnCOEJ4NVV9XLgNGB1kjOBjcD2qloBbG/PSXIynSUwTwFWA5clWdDe63JgA511lle01wHWA49U1UnAJcDFs7BvkqRpmDIQquMf2tNnt1sBa4Atrb4FOLs9XgNcU1VPVNW9wChwRpITgCOr6paqKuCqCWPG3+t6YNX40YMk6fDo6xxCkgVJbgf2ATdV1Q7g+KraC9Duj2ubLwEe6Bq+u9WWtMcT6weMqar9wKPAMT362JBkJMnI2NhYf3soSepLX4FQVU9W1WnAUjp/7Z86yea9/rKvSeqTjZnYx+aqWllVKxcvXjxV25KkaZjWt4yq6qvAJ+nM/T/YpoFo9/vaZruBE7uGLQX2tPrSHvUDxiRZCBwFPDyd3iRJB6efbxktTvKi9vgI4EeBLwHbgHVts3XADe3xNmBt++bQcjonj29t00qPJTmznR84b8KY8fc6B7i5nWeQJB0m/fz89QnAlvZNoWcBW6vqz5PcAmxNsh64HzgXoKp2JdkK3AHsBy6oqifbe50PXAkcAdzYbgBXAFcnGaVzZLB2NnZOktS/KQOhqr4AvKJH/SFg1dOM2QRs6lEfAZ5y/qGqHqcFiiRpMLxSWZIEGAiSpMZAkCQBBoIkqTEQJEmAgSBJagwESRJgIEiSGgNBkgQYCJKkxkCQJAEGgiSpMRAkSYCBIElqDARJEmAgSJIaA0GSBPS3pvKJSf4yyZ1JdiV5R6svSnJTkrvb/dFdYy5MMprkriRnddVPT7KzvXZpW1uZtv7yta2+I8my2d9VSdJk+jlC2A/8alV9J3AmcEGSk4GNwPaqWgFsb89pr60FTgFWA5e19ZgBLgc2ACvabXWrrwceqaqTgEuAi2dh3yRJ0zBlIFTV3qr6bHv8GHAnsARYA2xpm20Bzm6P1wDXVNUTVXUvMAqckeQE4MiquqWqCrhqwpjx97oeWDV+9CBJOjymdQ6hTeW8AtgBHF9Ve6ETGsBxbbMlwANdw3a32pL2eGL9gDFVtR94FDimx7+/IclIkpGxsbHptC5JmkLfgZDkhcAHgHdW1dcm27RHrSapTzbmwELV5qpaWVUrFy9ePFXLkqRp6CsQkjybThj8aVV9sJUfbNNAtPt9rb4bOLFr+FJgT6sv7VE/YEyShcBRwMPT3RlJ0sz18y2jAFcAd1bVf+96aRuwrj1eB9zQVV/bvjm0nM7J41vbtNJjSc5s73nehDHj73UOcHM7zyBJOkwW9rHNq4C3ADuT3N5qvwFcBGxNsh64HzgXoKp2JdkK3EHnG0oXVNWTbdz5wJXAEcCN7QadwLk6ySidI4O1B7lfkqRpmjIQquqv6D3HD7DqacZsAjb1qI8Ap/aoP04LFEnSYHilsiQJMBAkSY2BIEkCDARJUmMgSJIAA0GS1BgIkiTAQJAkNQaCJAkwECRJjYEgSQIMBElSYyBIkgADQZLUGAiSJMBAkCQ1/Syh+cdJ9iX5YldtUZKbktzd7o/ueu3CJKNJ7kpyVlf99CQ722uXtmU0aUttXtvqO5Ism91dlCT1o58lNK8E/idwVVdtI7C9qi5KsrE9//UkJ9NZ/vIU4MXAJ5J8e1tC83JgA/A3wEeB1XSW0FwPPFJVJyVZC1wM/NRs7Jw0Xyzb+JFBt9CX+y56/aBb0EGY8gihqj5NZ53jbmuALe3xFuDsrvo1VfVEVd0LjAJnJDkBOLKqbqmqohMuZ/d4r+uBVeNHD5Kkw2em5xCOr6q9AO3+uFZfAjzQtd3uVlvSHk+sHzCmqvYDjwLH9PpHk2xIMpJkZGxsbIatS5J6me2Tyr3+sq9J6pONeWqxanNVrayqlYsXL55hi5KkXmYaCA+2aSDa/b5W3w2c2LXdUmBPqy/tUT9gTJKFwFE8dYpKknSIzTQQtgHr2uN1wA1d9bXtm0PLgRXArW1a6bEkZ7bzA+dNGDP+XucAN7fzDJKkw2jKbxkleT/ww8CxSXYD7wYuArYmWQ/cD5wLUFW7kmwF7gD2Axe0bxgBnE/nG0tH0Pl20Y2tfgVwdZJROkcGa2dlzyRJ0zJlIFTVm57mpVVPs/0mYFOP+ghwao/647RAkSQNjlcqS5KA/i5Mk6R5xQv9evMIQZIEGAiSpMZAkCQBBoIkqTEQJEmAgSBJagwESRJgIEiSGgNBkgQYCJKkxkCQJAEGgiSpMRAkSYCBIElqDARJEjBEgZBkdZK7kowm2TjofiTpmWYoAiHJAuB/Aa8FTgbelOTkwXYlSc8sQxEIwBnAaFXdU1XfAK4B1gy4J0l6RklVDboHkpwDrK6qn2/P3wJ8T1X90oTtNgAb2tOXAXcd1kZn5ljgK4NuYh7x85w9fpaza658ni+pqsW9XhiWNZXTo/aUpKqqzcDmQ9/O7EkyUlUrB93HfOHnOXv8LGfXfPg8h2XKaDdwYtfzpcCeAfUiSc9IwxIIfwusSLI8yXOAtcC2AfckSc8oQzFlVFX7k/wS8HFgAfDHVbVrwG3Nljk1xTUH+HnOHj/L2TXnP8+hOKksSRq8YZkykiQNmIEgSQIMBElSYyBI0gwlObFH7V8NopfZYCAcAkm291PT5JI8K8kXB93HfJJke5LXTajN+W/HDNC9Sd6f5PldtY8OrJuDZCDMoiTPS7IIODbJ0UkWtdsy4MWD7W7uqap/AT6f5FsH3cs8shz49STv7qrN6atrB2wn8BngM0le2mq9fnlhThiK6xDmkbcC76TzH//PdtW/RufXXDV9JwC7ktwKfH28WFVvHFxLc9pXgVXApUk+DLx5wP3MdVVVlyX5PPDhJL9Oj5/dmSu8DuEQSPLLVfUHg+5jPkjyQ73qVfWpw93LfJDkc1X1ivb43wG/ChxdVUsH2tgcNeHzPAG4FlhZVc+ffORwMhBmUZJXV9XNSf5Nr9er6oOHu6f5IMlLgBVV9Yk2V7ugqh4bdF9zUZK3VtV7up6fDlxQVT83wLbmrCQnVNXerucLge+rqk8PsK0Zc8podv0QcDPw4z1eK8BAmKYkv0DnJ88XAS8FlgB/SGfaQ9NUVe9J8nLgB1rpM4bBQfnHJJfwzc/zU8BvD7Cfg+IRgoZaktvpLKC0o+vQfGdVfddgO5ubkrydTsCO/3HyE8BmpzhnJskHgC8CW1rpLcDLq6rnLMGwMxAOgSTv6lF+FLitqm4/3P3MZUl2VNX3jM/VtkPyz1bVdw+6t7koyReA762qr7fnLwBu8fOcmSS3V9VpU9XmCr92emisBN5GZ3pjCZ2/yH4YeG+SXxtgX3PRp5L8BnBEktcA1wEfHnBPc1mAJ7ueP8kc/prkEPinJN8//iTJq4B/GmA/B8UjhEMgyceBf1tV/9CevxC4ns7h+W1VdfIg+5tLkjwLWA/8GJ3/cH28qt472K7mrnb0ug74UCudDVxZVb8/uK7mrnY+5irgqFZ6BFhXVV8YXFczZyAcAknupDOP+I32/LnA7VX1nd1fU9PUkpxeVbdNqP14VXmUMENJXgl8P52A/XRVfW7ALc1ZXdPDL2z3/8Acnh72W0aHxp8Bf5Pkhvb8x4H3t/naOwbX1pz03iTrqmonQJI30bn4z0CYgXYl/X3tNl57dlX986B6muNWtts2OgH703RWgHxbkuuq6r8Msrnp8gjhEGnf7x7/K+yvqmpkwC3NSUm+jc5028/Q+TzPA95QVY8OtLE5Ksl9dNYvf4TO/zZfBOwF9gG/MPFoTJObb9PDnlQ+dI4AvtbmZr+cZPmgG5qLquoeOmtsfwA4B/gxw+CgfAx4XVUdW1XHAK8FtgK/CFw20M7mpm8FvtH1/J+Bl1TVPwFPDKalmXPK6BBoPxy2EngZ8D7g2cCfAK8aZF9zSZKdHPibMIvorLe9Iwl+TXLGVlbV28afVNVfJPndqnpXO9el6ZlX08MGwqHxE8AraD9wV1V7knzLYFuac94w6AbmqYfbD7Bd057/FPBIkgXAvwyurbmpqn4nyUf55vTw27qmh39mcJ3NjIFwaHyjqipJwf+/+EfTUFVfBkhyJrBr/LeLWrCeDHx5gO3NZT8NvBv433T+A/aZVlsA/OQA+5qz2nmXeXHuxZPKh0CSfw+sAF4D/Gfg54A/8+cBpi/J54BXVvsfarsuYaSqXjnYzuamJN/WzstIT2EgHCLtqtrui6luGnBLc9LT/DTAFzyHMDNJPk3n6vm/BT5N58ftdg62Kw0LA0FDLckHgU8Cl7fSLwI/UlVnD6ypOS7Jc4B/TefnVN4KvLCqFg20KQ0FA2EWJXmM3qslhc7KSkce5pbmvCTHAZcCr26lTwDvrKp9g+tq7mq/u/MD7fYi4HY6RwnvH2hjGgoGgvQMkuRJYITOua2Pjv+8igRemKYhl2Rpkg8l2ZfkwSQfSOJyjzN3DJ0FXL4X+FiSTyT5nQH3pCFhIGjYvY/O78S8mM7J0A+3mmagqr4K3APcS+cnK14K/OBAm9LQcMpIQ22+LUAyaEn+DrgL+Cs61yDscNpI47wwTcPuK0neDIyf9HwT8NAA+5nrVlSVVySrJ6eMNOx+js4VtH9PZ4rjnFbTzJyUZHuSLwIk+e4kvznopjQcnDKSnkGSfAr4D8B7xhdqSvLFqjp1sJ1pGDhlpKGU5A/ofU0HAFX19sPYznzy/Kq6NTlgGeX9g2pGw8VA0LByQaFD4ytJXkoL2yTn0JmKk5wy0tzQfuW0xlem0sy0Feg2A99HZ9W0e4GfGf91WT2zGQgaaklOBa6ms0BOgDHgvKraNdDG5qi2CM45wDI6n+nX6ATtbw+yLw0Hp4w07DYD76qqvwRI8sPAe+n8havpuwH4Kp3Fm/YMuBcNGQNBw+4F42EAUFWfdMGhg7K0qlYPugkNJ69D0LC7J8l/TLKs3X6Tzry3Zuavk3zXoJvQcPIcgoZakqOB36KzZi10FnX5T+03eTRNSe4ATqITqk/wzZ9md8EhGQgabknOrarrpqqpP0le0qvut4wEBoKGXJLPTlw/uVdN0sHzpLKGUpLXAq8DliS5tOulI/HKWumQMBA0rPbQuVr5jcBtXfXHgF8ZSEfSPOeUkYZakmfTOfH5HXR+buEuf79fOjQ8QtCwew3wHuDv6ATD8iRvraobB9uWNP94hKChluRLwBuqarQ9fynwkar6jsF2Js0/XpimYbdvPAyae4B9g2pGms88QtBQS3I58BJgK51zCOfSWRP4/wBU1QcH1500vxgIGmpJ3jfJy1VVLqcpzRIDQUMtyaKqenjQfUjPBJ5D0LDbkeS6JK/LhHUfJc0uA0HD7tvprInwFmA0ye8m+fYB9yTNS04Zac5I8iPAnwAvAD4PbKyqWwbblTR/GAgaakmOAd5M5wjhQeAKYBtwGnBdVS0fYHvSvOKVyhp2t9BZU/nsqtrdVR9J8ocD6kmalzxC0NBKsgD4r1X1rkH3Ij0TeFJZQ6uqngRePug+pGcKp4w07G5Psg24Dvj6eNErlKXZZyBo2C0CHgJe3VUrwECQZpmBoGH3LOAdVfVVgCRHA7832Jak+clzCBp23z0eBgBV9QjwigH2I81bBoKG3bPaUQHQ+W0jPLKVDgn/j6Vh93vAXye5ns65g58ENg22JWl+8joEDb0kJ9M5qRxge1XdMeCWpHnJQJAkAZ5DkCQ1BoIkCTAQJEmNgSBJAuD/ATTZN0P5BDfSAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_df[\"subclass\"].value_counts().plot.bar()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "A3A4-_SQSral"
   },
   "source": [
    "## Observations on `subclass` Catagorical Data\n",
    "\n",
    "For subclass there is a tail forming in the distribution, if accuracy in determining non `legit` subclasses is sub-par perhaps applying a resample technique such as bootstrapping in training may be required.\n",
    "\n",
    "Accuracy will not be a sufficient metric given the imbalances in representation across `subclass`. If we take the view of a type 2 error (False Positive) is more undesireable than a type 1 error (False Negative) we should look to optimise model specificity and precision over accuracy."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "pPG_3h9YUMtu"
   },
   "source": [
    "# Distibutions of Quantitative Data\n",
    "\n",
    "While there aren't any Quantitative fields in the dataset we can generate some simple values, such as length of string fields and occurances of characters in strings.\n",
    "\n",
    "Furthermore converting our string data to numeric data will faciliate further quality checks."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "5OBY8GJaVUMB"
   },
   "source": [
    "## Top Level Domain\n",
    "\n",
    "Let's split out the Top Level Domain (e.g. `com` from `google.com`) and so if there are any insights to be had."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "hZfUot2nU5sd"
   },
   "outputs": [],
   "source": [
    "data_df[\"tld\"] = data_df[\"host\"].apply(lambda x: \".\".join(x.split(\".\")[1:]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "xF4GIMWWE_D_"
   },
   "source": [
    "Let's look at the top 10 Top Level Domains."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 307
    },
    "colab_type": "code",
    "id": "zK9n6bVXVHcR",
    "outputId": "a0891e0e-ffb1-492b-ff79-6d17ea2975b6"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2091e83aa20>"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_df[\"tld\"].value_counts().head(10).plot.bar()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "qxXtQ9BAjgib"
   },
   "source": [
    "\n",
    "The number of of `com` TLDs is not that suprising as it is seen as a desirable domain for legitimate purposes.\n",
    "\n",
    "The the number of `ru` TLDs is both suprising and not suprising upon reflection of potential bias in the collection of the dataset.\n",
    "\n",
    "After the 4th most frequent TLD we have a long tailed distribution. \n",
    "We may not have enough observations if we stratify by `tld`, let's look at the interquartile ranges."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 170
    },
    "colab_type": "code",
    "id": "K1fT314fjXH8",
    "outputId": "57eef748-73a3-42aa-ba70-909c3e984e44"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count      723.000000\n",
       "mean       185.236515\n",
       "std       2377.449459\n",
       "min          1.000000\n",
       "25%          1.000000\n",
       "50%          2.000000\n",
       "75%         11.000000\n",
       "max      45070.000000\n",
       "Name: tld, dtype: float64"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[\"tld\"].value_counts().describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "oFtUWJ8diGf3"
   },
   "source": [
    "Let's group on TLD and class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 235
    },
    "colab_type": "code",
    "id": "_T7oz1PPlUH_",
    "outputId": "213fdbd6-358e-4a95-c23f-dcd5aa4d18fa"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>occurrences</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>tld</th>\n",
       "      <th>class</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>ab.ca</th>\n",
       "      <th>legit</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>abruzzo.it</th>\n",
       "      <th>legit</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ac</th>\n",
       "      <th>legit</th>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ac.at</th>\n",
       "      <th>legit</th>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ac.id</th>\n",
       "      <th>legit</th>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  occurrences\n",
       "tld        class             \n",
       "ab.ca      legit            1\n",
       "abruzzo.it legit            1\n",
       "ac         legit            4\n",
       "ac.at      legit            1\n",
       "ac.id      legit            5"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped_tld_class_df = data_df[[\"tld\", \"class\", \"domain\"]].groupby([\"tld\", \"class\"]).agg(\"count\")\n",
    "grouped_tld_class_df.columns = [\"occurrences\"]\n",
    "grouped_tld_class_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "jab0oQiSol41"
   },
   "source": [
    "The groupby operation will omit group combinations that do not exist, \n",
    "\n",
    "i.e. If there te combination of `za` lga and a `dga` class is not observed in the dataset then it will not appear in the grouped by dataset.\n",
    "\n",
    "For our purposes we do want to be able to query which combinations do not exist."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "-HyLyGlorhNc"
   },
   "source": [
    "Let's create an empty dataframe to join onto `grouped_tld_class_df` by using a multi-index built out of the product of `tld` and `class`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "WdCP2Rv_pQEE",
    "outputId": "2f188e4f-4334-4599-ba6a-dada968bf5f5"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>tld</th>\n",
       "      <th>class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>com.ar</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>com.ar</td>\n",
       "      <td>dga</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>com</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>com</td>\n",
       "      <td>dga</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>cn</td>\n",
       "      <td>legit</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      tld  class\n",
       "0  com.ar  legit\n",
       "1  com.ar    dga\n",
       "2     com  legit\n",
       "3     com    dga\n",
       "4      cn  legit"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tld_class_combinations = pd.MultiIndex.from_product([data_df[\"tld\"].unique(),data_df[\"class\"].unique()], names = [\"tld\", \"class\"])\n",
    "\n",
    "# simple check that we have the expected number of combinations\n",
    "assert(len(tld_class_combinations) == len(data_df[\"tld\"].unique()) * len(data_df[\"class\"].unique()))\n",
    "\n",
    "empty_tld_class_combinations_df = pd.DataFrame(index=tld_class_combinations).reset_index()\n",
    "empty_tld_class_combinations_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "LcNle0iDrxrL"
   },
   "source": [
    "We'll join them on using a outer join so that combinations not present in the `grouped_tld_class_df` will have a known empty value `np.nan` in a new column "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "qnED72JezBht",
    "outputId": "3b8e1f29-fa67-4e6f-b0d0-3265a4d8f459"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>tld</th>\n",
       "      <th>class</th>\n",
       "      <th>occurrences</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ab.ca</td>\n",
       "      <td>legit</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>abruzzo.it</td>\n",
       "      <td>legit</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ac</td>\n",
       "      <td>legit</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ac.at</td>\n",
       "      <td>legit</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ac.id</td>\n",
       "      <td>legit</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          tld  class  occurrences\n",
       "0       ab.ca  legit            1\n",
       "1  abruzzo.it  legit            1\n",
       "2          ac  legit            4\n",
       "3       ac.at  legit            1\n",
       "4       ac.id  legit            5"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped_tld_class_df.reset_index().head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "BmDWRjF2qULD"
   },
   "outputs": [],
   "source": [
    "exhaustive_tld_class_df = grouped_tld_class_df.reset_index().merge(\n",
    "    empty_tld_class_combinations_df, \n",
    "    how = \"outer\", \n",
    "    on = [\"tld\", \"class\"],\n",
    "    suffixes=(\"\", \"_empty\"), \n",
    "    ).fillna(np.nan).sort_values([\"tld\", \"class\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "jc4fca-Or-VP",
    "outputId": "af5d241c-cbbd-4eda-b078-e63bc49d5d4f"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>tld</th>\n",
       "      <th>class</th>\n",
       "      <th>occurrences</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1187</th>\n",
       "      <td>ab.ca</td>\n",
       "      <td>dga</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ab.ca</td>\n",
       "      <td>legit</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1367</th>\n",
       "      <td>abruzzo.it</td>\n",
       "      <td>dga</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>abruzzo.it</td>\n",
       "      <td>legit</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1006</th>\n",
       "      <td>ac</td>\n",
       "      <td>dga</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             tld  class  occurrences\n",
       "1187       ab.ca    dga          NaN\n",
       "0          ab.ca  legit          1.0\n",
       "1367  abruzzo.it    dga          NaN\n",
       "1     abruzzo.it  legit          1.0\n",
       "1006          ac    dga          NaN"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "exhaustive_tld_class_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "mKJ01k_PmltW"
   },
   "source": [
    "How many tld do not have any `legit` observations?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "bZ5F2fInww8A",
    "outputId": "9a49c703-67c4-41d6-bb60-79b3c94e1b06"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "missing_legit 0\n"
     ]
    }
   ],
   "source": [
    "missing_legit = exhaustive_tld_class_df[ \n",
    "                     (exhaustive_tld_class_df[\"occurrences\"].isnull()) & \n",
    "                     (exhaustive_tld_class_df[\"class\"] == \"legit\") \n",
    "                     ].shape[0]\n",
    "\n",
    "print(\"missing_legit\", missing_legit)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "P4lL8aZX8H-x",
    "outputId": "ad5cbf24-a3a8-4689-8482-09f8acc5a6c7"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "missing_legit / len(exhaustive_tld_class_df[\"tld\"].unique())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Z-W8IZfPwVNN"
   },
   "source": [
    "How about tld without a `dga` observation?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "BCnReghYxJru",
    "outputId": "26849eb8-955d-4b5a-d2dd-b0609c779024"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "missing_dga 718\n"
     ]
    }
   ],
   "source": [
    "missing_dga = exhaustive_tld_class_df[ \n",
    "                     (exhaustive_tld_class_df[\"occurrences\"].isnull()) & \n",
    "                     (exhaustive_tld_class_df[\"class\"] == \"dga\") \n",
    "                     ].shape[0]\n",
    "                     \n",
    "print(\"missing_dga\", missing_dga)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "8ylAy9zc8bzQ"
   },
   "source": [
    "718 Top Level Domains do not have a recorded observation of a `dga` class.\n",
    "\n",
    "Let's express that as a ratio.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "Z7qDFroj7xLB",
    "outputId": "e9fe6ed7-9137-4d5a-a9c7-3f779806c1e7"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9930843706777317"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "missing_dga / len(exhaustive_tld_class_df[\"tld\"].unique())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "nQwgg-vD8f1u"
   },
   "source": [
    "So which TLD have dga observations?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "Me_NCcjV8Spt",
    "outputId": "1ee287da-37c8-4d70-c660-e9bdccc9c5b0"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>tld</th>\n",
       "      <th>class</th>\n",
       "      <th>occurrences</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>biz</td>\n",
       "      <td>dga</td>\n",
       "      <td>1834.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>133</th>\n",
       "      <td>com</td>\n",
       "      <td>dga</td>\n",
       "      <td>2749.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>481</th>\n",
       "      <td>net</td>\n",
       "      <td>dga</td>\n",
       "      <td>3666.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>535</th>\n",
       "      <td>org</td>\n",
       "      <td>dga</td>\n",
       "      <td>2750.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613</th>\n",
       "      <td>ru</td>\n",
       "      <td>dga</td>\n",
       "      <td>41666.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     tld class  occurrences\n",
       "44   biz   dga       1834.0\n",
       "133  com   dga       2749.0\n",
       "481  net   dga       3666.0\n",
       "535  org   dga       2750.0\n",
       "613   ru   dga      41666.0"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "exhaustive_tld_class_df[ \n",
    "                     (~exhaustive_tld_class_df[\"occurrences\"].isnull()) & \n",
    "                     (exhaustive_tld_class_df[\"class\"] == \"dga\") \n",
    "                     ]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "SPGPW63AxTdE"
   },
   "source": [
    "Let's visualise the absolute difference between classes for our most frequently appearing `tld`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 326
    },
    "colab_type": "code",
    "id": "SV-FdG5i3gGi",
    "outputId": "12c2d3d8-2ef8-49a8-88f4-52c30016b915"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2090431dcc0>"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "abs_difference_tld_class_df = exhaustive_tld_class_df.fillna(0).copy()\n",
    "abs_difference_tld_class_df.loc[abs_difference_tld_class_df[\"class\"] == \"legit\", \"occurrences\"] *= -1\n",
    "abs_difference_tld_class_df = pd.DataFrame(abs_difference_tld_class_df.groupby(\"tld\")[\"occurrences\"].sum().abs()).sort_values(\"occurrences\", ascending = False)\n",
    "abs_difference_tld_class_df.columns = [\"abs_delta_occurrences\"]\n",
    "abs_difference_tld_class_df.head(30).plot.bar()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "JYK2zg47kMe5"
   },
   "source": [
    "Let's look at the breakdown of classess for the top two most frequent `tld`: `com` and `ru`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 142
    },
    "colab_type": "code",
    "id": "fwQxH7lBkIJO",
    "outputId": "33972463-e6a5-425a-eb37-265bf81454ed"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>occurrences</th>\n",
       "      <th>ratio</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>class</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>dga</th>\n",
       "      <td>2749</td>\n",
       "      <td>0.060994</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>42321</td>\n",
       "      <td>0.939006</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       occurrences     ratio\n",
       "class                       \n",
       "dga           2749  0.060994\n",
       "legit        42321  0.939006"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "com_by_class_df = grouped_tld_class_df.loc[\"com\", :]\n",
    "com_by_class_df[\"ratio\"] = com_by_class_df[\"occurrences\"] / com_by_class_df[\"occurrences\"].sum()\n",
    "com_by_class_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 142
    },
    "colab_type": "code",
    "id": "YZemPnEAioYA",
    "outputId": "44ee6adb-1995-4c7a-dec0-e63f8f42293c"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>occurrences</th>\n",
       "      <th>ratio</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>class</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>dga</th>\n",
       "      <td>41666</td>\n",
       "      <td>0.942904</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>2523</td>\n",
       "      <td>0.057096</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       occurrences     ratio\n",
       "class                       \n",
       "dga          41666  0.942904\n",
       "legit         2523  0.057096"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ru_by_class_df = grouped_tld_class_df.loc[\"ru\", :]\n",
    "ru_by_class_df[\"ratio\"] = ru_by_class_df[\"occurrences\"] / ru_by_class_df[\"occurrences\"].sum()\n",
    "ru_by_class_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "EMDdbocEVbbV"
   },
   "source": [
    "## Observations on Top Level Domain\n",
    "\n",
    "Given: \n",
    "* We do not have exhaustive observations for the product of `tld` and `class` ~99% of TLD do not have observations where class is `dga`\n",
    "* The imbalance where there are observations for both `classes` is heavily skewed\n",
    "\n",
    "Thus `tld` as a feature may be difficult to generalise."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "5Tqz3S2j9fpf"
   },
   "source": [
    "## Length of Domain\n",
    "\n",
    "Lets review if the length of a `domain` has any indication on the `class` of an observation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "IU_-bvLy9onh",
    "outputId": "39354509-2cb5-470e-ee63-7938f4864844"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(114525, 3)"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domains_df = data_df[[\"domain\", \"class\", \"subclass\"]].drop_duplicates()\n",
    "domains_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "7OdRlzq09vDu"
   },
   "outputs": [],
   "source": [
    "domains_df[\"domain_len\"] = domains_df[\"domain\"].apply(len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 142
    },
    "colab_type": "code",
    "id": "AGlnYUjt96JG",
    "outputId": "6a4ffd07-9c28-427d-f172-2daef578a9ed"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>class</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>dga</th>\n",
       "      <td>52665.0</td>\n",
       "      <td>17.726764</td>\n",
       "      <td>5.946592</td>\n",
       "      <td>12.0</td>\n",
       "      <td>13.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>32.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>61860.0</td>\n",
       "      <td>10.761752</td>\n",
       "      <td>3.300436</td>\n",
       "      <td>7.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>56.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         count       mean       std   min   25%   50%   75%   max\n",
       "class                                                            \n",
       "dga    52665.0  17.726764  5.946592  12.0  13.0  15.0  25.0  32.0\n",
       "legit  61860.0  10.761752  3.300436   7.0   8.0  10.0  12.0  56.0"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domains_len_stats_df = domains_df.groupby(\"class\")[\"domain_len\"].describe()\n",
    "domains_len_stats_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 333
    },
    "colab_type": "code",
    "id": "vcdGtI4_ZdZ3",
    "outputId": "970274f0-3581-4bd1-e797-175901ca0493"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "class\n",
       "dga      AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "legit    AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "Name: domain_len, dtype: object"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "domains_df.groupby(\"class\")[\"domain_len\"].plot.kde(legend=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "0WXGNyIH-Y8h"
   },
   "source": [
    "Looking a the `min` value there is something VERY interesting."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "2Hy7rZbYAeoc"
   },
   "source": [
    "Let's review the interquartile values to get a better appreciation of the standard deviation values above."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 308
    },
    "colab_type": "code",
    "id": "iBYKQosUAPpg",
    "outputId": "31b7c5e1-8312-49b3-8159-9199dae73b49"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2090ffd6048>"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "domains_len_stats_df[domains_len_stats_df.columns[4:]].plot.bar()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "20kXFiodA4Q7"
   },
   "source": [
    "Let's have a look at these really long domains."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 607
    },
    "colab_type": "code",
    "id": "C39Iu40PA_0A",
    "outputId": "6b75f21e-dc24-4011-f9de-5a74deaa1f7c"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>domain</th>\n",
       "      <th>domain_len</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>class</th>\n",
       "      <th>subclass</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"6\" valign=\"top\">dga</th>\n",
       "      <th>goz</th>\n",
       "      <td>fahqfqjnztfukbrsrsgerwqctkdeusjv</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>goz</th>\n",
       "      <td>lbpzkvfmemaiaqhyemijamofdqbuqkmj</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>goz</th>\n",
       "      <td>onwgbaaenvofifsoaqnvqokzfqlbmzxc</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>goz</th>\n",
       "      <td>ozuwydmfvkrwydgyhmmjfifmovnbrofu</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>goz</th>\n",
       "      <td>twqojncqtsfikbscqsaunfpjcakztkxo</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>goz</th>\n",
       "      <td>ydeudipztwgaaupvrkdqucbenfoncady</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"11\" valign=\"top\">legit</th>\n",
       "      <th>legit</th>\n",
       "      <td>americansforresponsiblesolutions</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>b117f8da23446a91387efea0e428392a</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>bipolardisorderdepressionanxiety</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>deutsche-wirtschafts-nachrichten</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>ostsaechsische-sparkasse-dresden</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>dellbackupandrecoverycloudstorage</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>deutsche-mittelstands-nachrichten</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>insearchoftheworldsmostbeautifulwoman</td>\n",
       "      <td>37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>sparkasse-oberlausitz-niederschlesien</td>\n",
       "      <td>37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>article-directory-free-submission-free-content</td>\n",
       "      <td>46</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>gay-sex-pics-porn-pictures-gay-sex-porn-gay-se...</td>\n",
       "      <td>56</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                           domain  domain_len\n",
       "class subclass                                                               \n",
       "dga   goz                        fahqfqjnztfukbrsrsgerwqctkdeusjv          32\n",
       "      goz                        lbpzkvfmemaiaqhyemijamofdqbuqkmj          32\n",
       "      goz                        onwgbaaenvofifsoaqnvqokzfqlbmzxc          32\n",
       "      goz                        ozuwydmfvkrwydgyhmmjfifmovnbrofu          32\n",
       "      goz                        twqojncqtsfikbscqsaunfpjcakztkxo          32\n",
       "      goz                        ydeudipztwgaaupvrkdqucbenfoncady          32\n",
       "legit legit                      americansforresponsiblesolutions          32\n",
       "      legit                      b117f8da23446a91387efea0e428392a          32\n",
       "      legit                      bipolardisorderdepressionanxiety          32\n",
       "      legit                      deutsche-wirtschafts-nachrichten          32\n",
       "      legit                      ostsaechsische-sparkasse-dresden          32\n",
       "      legit                     dellbackupandrecoverycloudstorage          33\n",
       "      legit                     deutsche-mittelstands-nachrichten          33\n",
       "      legit                 insearchoftheworldsmostbeautifulwoman          37\n",
       "      legit                 sparkasse-oberlausitz-niederschlesien          37\n",
       "      legit        article-directory-free-submission-free-content          46\n",
       "      legit     gay-sex-pics-porn-pictures-gay-sex-porn-gay-se...          56"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domains_df[domains_df[\"domain_len\"] >= domains_len_stats_df[\"max\"].min()].sort_values([\"class\", \"subclass\", \"domain_len\"]).set_index([\"class\", \"subclass\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "CD-qe0CdBaHZ"
   },
   "source": [
    "The records that have the largest `domain_len` have repeating \"words\" that can be identified due to hyphen usuage. This might be a pattern that is repeated elsewhere. However it is doubtful that we could reliably generate a measure of \"unique words\" without implementing a dictionary which would be a considerable effort.\n",
    "\n",
    "The other interesting quirk is the presence of `b117f8da23446a91387efea0e428392a` as a legit domain, it looks very much like a `dga` - is this perhaps a mislabeled observation?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "uN3hS8kYj1Ds"
   },
   "source": [
    "Let's count the number of records that have a `domain_len` greater than the 75th percentile accross the classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 111
    },
    "colab_type": "code",
    "id": "0SBvBj5yBWRD",
    "outputId": "b415ca4e-d8a0-4dd0-d38e-92d85f4ecd5e"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>dga</th>\n",
       "      <td>14643</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>124</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       class\n",
       "dga    14643\n",
       "legit    124"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "long_domains_by_class_df = pd.DataFrame(domains_df[domains_df[\"domain_len\"] >= domains_len_stats_df[\"75%\"].max()][\"class\"].value_counts())\n",
    "long_domains_by_class_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "gji5b4RlCs6U"
   },
   "source": [
    "These long `legit` domains, what are they as a ratio of all legitimate domains in the dataset. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "r97SVokKCjo0",
    "outputId": "3087ad02-974c-41f6-d6e6-193517de4716"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.002004526349822179"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "long_domains_by_class_df.loc[\"legit\"][\"class\"] / domains_len_stats_df.loc[\"legit\"][\"count\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Zhe07Ect_dfC"
   },
   "source": [
    "## Observations on Length of Domain\n",
    "\n",
    "For this we can deduce that in the dataset we've been given a `dga` record will have at a mimimum a domain of 12 characters long.\n",
    "\n",
    "As for the maximum length where a record is `legit` having a `domain_len` >= 25 is the exception. Potentially we could discard these records, given they represent < 1% of all legit?\n",
    "\n",
    "However there is more variation in the length of `domain` for `dga` records. But the fact there is a point of clear separation (< 12 characters in length), the length of the `domain` would be a useful feature.\n",
    "\n",
    "There is a danger that `domain_len` might have undue importance due to the minimum length of 12 for `dga` - this is something that will need to be considered in model evaluation."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ysokwfv4DJEm"
   },
   "source": [
    "## Vowels\n",
    "\n",
    "It would seem that vowels are common occurance across languages as per [Vowel Systems](https://en.wikipedia.org/wiki/Vowel#Systems) on Wikipedia.\n",
    "\n",
    "There are some notable exceptions:\n",
    "* `y` is considered a vowel in certain conditions\n",
    "* `r` as a vowel in Czech and Slovak\n",
    "* Danish has 13 vowels: (a) ɑ (ɐ) e ə ɛ i o ɔ u ø œ y\n",
    "\n",
    "However given that DNS servers resolve non-ASCII characters into encoded ASCII characters we may find that all characters our dataset fall within the standard ASCII ordinal values.\n",
    "If this holds true then we can generate feature on the presence of `['a','e','i','o','u','y']`.\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "cxboJseLLwWH"
   },
   "source": [
    "\n",
    "### Vowel Counts\n",
    "Let's generate a count for the vowels."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "EnfekscWNIMt"
   },
   "outputs": [],
   "source": [
    "vowels = [\"a\", \"e\", \"i\", \"o\", \"u\", \"y\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "SJPAOlMqUj2R"
   },
   "source": [
    "Let's get the actual vowel counts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "pHUv9iF7U_le"
   },
   "outputs": [],
   "source": [
    "domain_letter_split = data_df[\"domain\"].apply(lambda x: [w for w in x])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "sZMoryL6UmU7"
   },
   "outputs": [],
   "source": [
    "data_df[\"domain_vowel_count\"] = domain_letter_split.apply(lambda x: len([w for w in x if w in vowels]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 170
    },
    "colab_type": "code",
    "id": "ik3jYTuvUy7D",
    "outputId": "7fe9085f-89a1-4676-d6ca-10dafadaeebb"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    133926.000000\n",
       "mean          3.959873\n",
       "std           1.730825\n",
       "min           0.000000\n",
       "25%           3.000000\n",
       "50%           4.000000\n",
       "75%           5.000000\n",
       "max          18.000000\n",
       "Name: domain_vowel_count, dtype: float64"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[\"domain_vowel_count\"].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 142
    },
    "colab_type": "code",
    "id": "WRKpEKBGVTBW",
    "outputId": "f80eb229-bd1f-4ff2-fa40-bc428d1f34a8"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>class</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>dga</th>\n",
       "      <td>52665.0</td>\n",
       "      <td>3.733960</td>\n",
       "      <td>1.929190</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>16.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>81261.0</td>\n",
       "      <td>4.106287</td>\n",
       "      <td>1.571861</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>18.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         count      mean       std  min  25%  50%  75%   max\n",
       "class                                                       \n",
       "dga    52665.0  3.733960  1.929190  0.0  2.0  4.0  5.0  16.0\n",
       "legit  81261.0  4.106287  1.571861  0.0  3.0  4.0  5.0  18.0"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domain_by_class_vowel_stats_df = data_df[[\"class\", \"domain_vowel_count\"]].groupby(\"class\")[\"domain_vowel_count\"].describe()\n",
    "domain_by_class_vowel_stats_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 308
    },
    "colab_type": "code",
    "id": "zuM7ONydWBoz",
    "outputId": "3c7b7528-0ff7-40c1-fd2d-2f5dd37be63c"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x20911d2f5c0>"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "domain_by_class_vowel_stats_df[domain_by_class_vowel_stats_df.columns[1:]].plot.bar()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 333
    },
    "colab_type": "code",
    "id": "wLQ8lDJAiOEV",
    "outputId": "c910c896-1e72-40e4-b7aa-93d77b490cd9"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "class\n",
       "dga      AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "legit    AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "Name: domain_vowel_count, dtype: object"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_df[[\"class\", \"domain_vowel_count\"]].groupby(\"class\")[\"domain_vowel_count\"].plot.kde(\n",
    "    legend = True,\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "GO_1eIX1Vzzq"
   },
   "source": [
    "Not much of a difference in the values between classes. Perhaps if we calculated the number of vowels as a ratio of length of the domain we might get better insights."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "InMJWtiDX9hX"
   },
   "source": [
    "### Observations on Vowel Count\n",
    "\n",
    "* Raw vowel count does not provide a clear boundary between `class`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Vr6SSFJJfihq"
   },
   "source": [
    "### Vowel Ratios"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "GZgIXBoPVn_w"
   },
   "outputs": [],
   "source": [
    "data_df[\"domain_vowel_ratio\"]  = data_df[\"domain_vowel_count\"] / data_df[\"domain\"].apply(len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 142
    },
    "colab_type": "code",
    "id": "ZycoUX3t4PZb",
    "outputId": "063bbc50-1d6e-4df9-9706-422a26a715b2"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>class</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>dga</th>\n",
       "      <td>52665.0</td>\n",
       "      <td>0.219425</td>\n",
       "      <td>0.108766</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.142857</td>\n",
       "      <td>0.214286</td>\n",
       "      <td>0.285714</td>\n",
       "      <td>0.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>81261.0</td>\n",
       "      <td>0.392103</td>\n",
       "      <td>0.095280</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.400000</td>\n",
       "      <td>0.444444</td>\n",
       "      <td>0.9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         count      mean       std  min       25%       50%       75%  max\n",
       "class                                                                     \n",
       "dga    52665.0  0.219425  0.108766  0.0  0.142857  0.214286  0.285714  0.8\n",
       "legit  81261.0  0.392103  0.095280  0.0  0.333333  0.400000  0.444444  0.9"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[[\"class\", \"domain_vowel_ratio\"]].groupby(\"class\")[\"domain_vowel_ratio\"].describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "cZ_q_QHae-gB"
   },
   "source": [
    "Let's visualise the overlap"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 333
    },
    "colab_type": "code",
    "id": "MsQaXepUdUwK",
    "outputId": "9c33b252-ad5d-4d7c-bc70-c84fb0514b8d"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "class\n",
       "dga      AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "legit    AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "Name: domain_vowel_ratio, dtype: object"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_df[[\"class\", \"domain_vowel_ratio\"]].groupby(\"class\")[\"domain_vowel_ratio\"].plot.kde(\n",
    "    legend = True,\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 142
    },
    "colab_type": "code",
    "id": "fhmZ8R4Z3N0I",
    "outputId": "0b286ebf-b48d-4259-fb94-ee529ba1fda5"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>class</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>dga</th>\n",
       "      <td>52665.0</td>\n",
       "      <td>0.219425</td>\n",
       "      <td>0.108766</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.142857</td>\n",
       "      <td>0.214286</td>\n",
       "      <td>0.285714</td>\n",
       "      <td>0.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>81261.0</td>\n",
       "      <td>0.392103</td>\n",
       "      <td>0.095280</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.400000</td>\n",
       "      <td>0.444444</td>\n",
       "      <td>0.9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         count      mean       std  min       25%       50%       75%  max\n",
       "class                                                                     \n",
       "dga    52665.0  0.219425  0.108766  0.0  0.142857  0.214286  0.285714  0.8\n",
       "legit  81261.0  0.392103  0.095280  0.0  0.333333  0.400000  0.444444  0.9"
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[[\"class\", \"domain_vowel_ratio\"]].groupby(\"class\")[\"domain_vowel_ratio\"].describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "7Ace-V-UfI4t"
   },
   "source": [
    "Lets see where how it plays out on IQR by class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 142
    },
    "colab_type": "code",
    "id": "nza_ATCFWTOh",
    "outputId": "a6eda8de-f238-4efe-efd6-7a7fee1fa52c"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>class</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>dga</th>\n",
       "      <td>52665.0</td>\n",
       "      <td>0.219425</td>\n",
       "      <td>0.108766</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.142857</td>\n",
       "      <td>0.214286</td>\n",
       "      <td>0.285714</td>\n",
       "      <td>0.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>81261.0</td>\n",
       "      <td>0.392103</td>\n",
       "      <td>0.095280</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.400000</td>\n",
       "      <td>0.444444</td>\n",
       "      <td>0.9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         count      mean       std  min       25%       50%       75%  max\n",
       "class                                                                     \n",
       "dga    52665.0  0.219425  0.108766  0.0  0.142857  0.214286  0.285714  0.8\n",
       "legit  81261.0  0.392103  0.095280  0.0  0.333333  0.400000  0.444444  0.9"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domain_by_class_vowel_ratio_stats_df = data_df[[\"class\", \"domain_vowel_ratio\"]].groupby(\"class\")[\"domain_vowel_ratio\"].describe()\n",
    "domain_by_class_vowel_ratio_stats_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "NjjHoj39fRWE"
   },
   "source": [
    "Given the 75th percentile of `domain_vowel_ratio` for `dga` is less than the 25th percentile for `legit` what is the overlap as a ratio?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "nRTccBCmeiAl",
    "outputId": "ae809009-5c94-4136-dbcc-c67a9bb1e7d3"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.17640688645229569"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "number_of_dga_domain_vowel_ratio_overlap = len(data_df[(data_df[\"class\"] == \"dga\") & (data_df[\"domain_vowel_ratio\"] >= domain_by_class_vowel_ratio_stats_df.loc[\"dga\"][\"75%\"])])\n",
    "number_of_dga_domain_vowel_ratio_overlap / len(data_df[(data_df[\"class\"] != \"dga\") & (data_df[\"domain_vowel_ratio\"] >= 0)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 308
    },
    "colab_type": "code",
    "id": "3nKa3na2f51Q",
    "outputId": "34f27ec9-3e2f-4817-af5d-72f25138bef1"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x209124ca710>"
      ]
     },
     "execution_count": 78,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "domain_by_class_vowel_ratio_stats_df[domain_by_class_vowel_ratio_stats_df.columns[1:]].plot.bar()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "AJBaBB48Wp7q"
   },
   "source": [
    "It would seem that `dga` class domains have fewer vowels in them as a ratio of all other letters.\n",
    "\n",
    "This is potentially a useful feature for distinguishing, as the 25th percentile `domain_vowel_ratio` for `legit` is higher than the 75th percential `domain_vowel_ratio` for `dga`.\n",
    "\n",
    "Although both have a minimum value of zero."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "ZDXbi5MfW1t3",
    "outputId": "6fe9a7e0-9baa-4618-b35c-3a8b8ea43360"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>host</th>\n",
       "      <th>domain</th>\n",
       "      <th>class</th>\n",
       "      <th>subclass</th>\n",
       "      <th>tld</th>\n",
       "      <th>domain_vowel_count</th>\n",
       "      <th>domain_vowel_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>25925</th>\n",
       "      <td>clktrkng.com</td>\n",
       "      <td>clktrkng</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>com</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>0594666.com</td>\n",
       "      <td>0594666</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>com</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28822</th>\n",
       "      <td>ctctcdn.com</td>\n",
       "      <td>ctctcdn</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>com</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19050</th>\n",
       "      <td>blt17688.com</td>\n",
       "      <td>blt17688</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>com</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>481</th>\n",
       "      <td>12333sh.gov.cn</td>\n",
       "      <td>12333sh</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>gov.cn</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 host    domain  class subclass     tld  domain_vowel_count  \\\n",
       "25925    clktrkng.com  clktrkng  legit    legit     com                   0   \n",
       "18        0594666.com   0594666  legit    legit     com                   0   \n",
       "28822     ctctcdn.com   ctctcdn  legit    legit     com                   0   \n",
       "19050    blt17688.com  blt17688  legit    legit     com                   0   \n",
       "481    12333sh.gov.cn   12333sh  legit    legit  gov.cn                   0   \n",
       "\n",
       "       domain_vowel_ratio  \n",
       "25925                 0.0  \n",
       "18                    0.0  \n",
       "28822                 0.0  \n",
       "19050                 0.0  \n",
       "481                   0.0  "
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[(data_df[\"domain_vowel_ratio\"] == 0) & (data_df[\"class\"] == \"legit\")].sample(5, random_state = 42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 224
    },
    "colab_type": "code",
    "id": "UUVshCDvXZ0G",
    "outputId": "271039c6-1a34-4abf-f71d-e942b446e860"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>host</th>\n",
       "      <th>domain</th>\n",
       "      <th>class</th>\n",
       "      <th>subclass</th>\n",
       "      <th>tld</th>\n",
       "      <th>domain_vowel_count</th>\n",
       "      <th>domain_vowel_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>23930</th>\n",
       "      <td>cgjrrcjmhncsdg.ru</td>\n",
       "      <td>cgjrrcjmhncsdg</td>\n",
       "      <td>dga</td>\n",
       "      <td>cryptolocker</td>\n",
       "      <td>ru</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81955</th>\n",
       "      <td>nnpwfdspvwvpkk.ru</td>\n",
       "      <td>nnpwfdspvwvpkk</td>\n",
       "      <td>dga</td>\n",
       "      <td>cryptolocker</td>\n",
       "      <td>ru</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31572</th>\n",
       "      <td>dgkhdkrdpwnkqh.ru</td>\n",
       "      <td>dgkhdkrdpwnkqh</td>\n",
       "      <td>dga</td>\n",
       "      <td>cryptolocker</td>\n",
       "      <td>ru</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3142</th>\n",
       "      <td>1hb8cs91x80rctvsxtp91ks942n.net</td>\n",
       "      <td>1hb8cs91x80rctvsxtp91ks942n</td>\n",
       "      <td>dga</td>\n",
       "      <td>newgoz</td>\n",
       "      <td>net</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>90143</th>\n",
       "      <td>plcsrmvbrpkbs.ru</td>\n",
       "      <td>plcsrmvbrpkbs</td>\n",
       "      <td>dga</td>\n",
       "      <td>cryptolocker</td>\n",
       "      <td>ru</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  host                       domain class  \\\n",
       "23930                cgjrrcjmhncsdg.ru               cgjrrcjmhncsdg   dga   \n",
       "81955                nnpwfdspvwvpkk.ru               nnpwfdspvwvpkk   dga   \n",
       "31572                dgkhdkrdpwnkqh.ru               dgkhdkrdpwnkqh   dga   \n",
       "3142   1hb8cs91x80rctvsxtp91ks942n.net  1hb8cs91x80rctvsxtp91ks942n   dga   \n",
       "90143                 plcsrmvbrpkbs.ru                plcsrmvbrpkbs   dga   \n",
       "\n",
       "           subclass  tld  domain_vowel_count  domain_vowel_ratio  \n",
       "23930  cryptolocker   ru                   0                 0.0  \n",
       "81955  cryptolocker   ru                   0                 0.0  \n",
       "31572  cryptolocker   ru                   0                 0.0  \n",
       "3142         newgoz  net                   0                 0.0  \n",
       "90143  cryptolocker   ru                   0                 0.0  "
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[(data_df[\"domain_vowel_ratio\"] == 0) & (data_df[\"class\"] != \"legit\")].sample(5, random_state = 42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "t1L1l5M3XLTi"
   },
   "source": [
    "Seems that the `legit` domains that have a zero vowel count tend to have digits in them, whereas the `dga` domains do not.\n",
    "\n",
    "Perhaps a similar measure for digits as a ratio of domain len would be beneficial."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "CaC19ngjY8Gx"
   },
   "source": [
    "Before we look into digits lets review the high `domain_vowel_ratio` records."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "kL6B93H_c_oA"
   },
   "outputs": [],
   "source": [
    "domain_vowel_ratio_75th_percentile_df = data_df[data_df[\"domain_vowel_ratio\"] >= domain_by_class_vowel_ratio_stats_df[\"75%\"].min()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "JgULZclSXJ_7",
    "outputId": "b428def8-a412-4f93-e890-48c4f3fa13c3"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>host</th>\n",
       "      <th>domain</th>\n",
       "      <th>class</th>\n",
       "      <th>subclass</th>\n",
       "      <th>tld</th>\n",
       "      <th>domain_vowel_count</th>\n",
       "      <th>domain_vowel_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>58155</th>\n",
       "      <td>inrixmedia.com</td>\n",
       "      <td>inrixmedia</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>com</td>\n",
       "      <td>5</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55681</th>\n",
       "      <td>iaosfbbjduor.ru</td>\n",
       "      <td>iaosfbbjduor</td>\n",
       "      <td>dga</td>\n",
       "      <td>cryptolocker</td>\n",
       "      <td>ru</td>\n",
       "      <td>5</td>\n",
       "      <td>0.416667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46101</th>\n",
       "      <td>gameduell.fr</td>\n",
       "      <td>gameduell</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>fr</td>\n",
       "      <td>4</td>\n",
       "      <td>0.444444</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>87534</th>\n",
       "      <td>patoghkade.com</td>\n",
       "      <td>patoghkade</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>com</td>\n",
       "      <td>4</td>\n",
       "      <td>0.400000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44644</th>\n",
       "      <td>freewebsiteadd.com</td>\n",
       "      <td>freewebsiteadd</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>com</td>\n",
       "      <td>6</td>\n",
       "      <td>0.428571</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     host          domain  class      subclass  tld  \\\n",
       "58155      inrixmedia.com      inrixmedia  legit         legit  com   \n",
       "55681     iaosfbbjduor.ru    iaosfbbjduor    dga  cryptolocker   ru   \n",
       "46101        gameduell.fr       gameduell  legit         legit   fr   \n",
       "87534      patoghkade.com      patoghkade  legit         legit  com   \n",
       "44644  freewebsiteadd.com  freewebsiteadd  legit         legit  com   \n",
       "\n",
       "       domain_vowel_count  domain_vowel_ratio  \n",
       "58155                   5            0.500000  \n",
       "55681                   5            0.416667  \n",
       "46101                   4            0.444444  \n",
       "87534                   4            0.400000  \n",
       "44644                   6            0.428571  "
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domain_vowel_ratio_75th_percentile_df.sample(5, random_state = 42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "BOcUsoEKaSVB",
    "outputId": "e28ac008-2aa6-4fb8-a7a1-d39159d55c62"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6515911772172692"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(domain_vowel_ratio_75th_percentile_df) / len(data_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 294
    },
    "colab_type": "code",
    "id": "XKYVCEZBcznE",
    "outputId": "5efd2349-a951-447c-c257-01cded2bc28c"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x20911edd4e0>"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_df[data_df[\"domain_vowel_ratio\"] >= domain_by_class_vowel_ratio_stats_df[\"75%\"].min()][\"class\"].value_counts().plot.bar()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "J4ftf0K8aHhe",
    "outputId": "b182611d-279e-43ed-e453-371aa5cc189e"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>host</th>\n",
       "      <th>domain</th>\n",
       "      <th>class</th>\n",
       "      <th>subclass</th>\n",
       "      <th>tld</th>\n",
       "      <th>domain_vowel_count</th>\n",
       "      <th>domain_vowel_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>132166</th>\n",
       "      <td>youyuan.com</td>\n",
       "      <td>youyuan</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>com</td>\n",
       "      <td>6</td>\n",
       "      <td>0.857143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130553</th>\n",
       "      <td>yahooeu.ru</td>\n",
       "      <td>yahooeu</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>ru</td>\n",
       "      <td>6</td>\n",
       "      <td>0.857143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7998</th>\n",
       "      <td>aaaauto.cz</td>\n",
       "      <td>aaaauto</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>cz</td>\n",
       "      <td>6</td>\n",
       "      <td>0.857143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>86254</th>\n",
       "      <td>ouyaoxiazai.com</td>\n",
       "      <td>ouyaoxiazai</td>\n",
       "      <td>legit</td>\n",
       "      <td>legit</td>\n",
       "      <td>com</td>\n",
       "      <td>9</td>\n",
       "      <td>0.818182</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42306</th>\n",
       "      <td>fiyeoyoeuijayag.ru</td>\n",
       "      <td>fiyeoyoeuijayag</td>\n",
       "      <td>dga</td>\n",
       "      <td>cryptolocker</td>\n",
       "      <td>ru</td>\n",
       "      <td>12</td>\n",
       "      <td>0.800000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                      host           domain  class      subclass  tld  \\\n",
       "132166         youyuan.com          youyuan  legit         legit  com   \n",
       "130553          yahooeu.ru          yahooeu  legit         legit   ru   \n",
       "7998            aaaauto.cz          aaaauto  legit         legit   cz   \n",
       "86254      ouyaoxiazai.com      ouyaoxiazai  legit         legit  com   \n",
       "42306   fiyeoyoeuijayag.ru  fiyeoyoeuijayag    dga  cryptolocker   ru   \n",
       "\n",
       "        domain_vowel_count  domain_vowel_ratio  \n",
       "132166                   6            0.857143  \n",
       "130553                   6            0.857143  \n",
       "7998                     6            0.857143  \n",
       "86254                    9            0.818182  \n",
       "42306                   12            0.800000  "
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[data_df[\"domain_vowel_ratio\"] >= domain_by_class_vowel_ratio_stats_df[\"max\"].min()].sample(5, random_state = 42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "x8UpcuM2Y7Qu"
   },
   "source": [
    "Looking at the above a potential alternative would be to calculate the ratios based on distinct character counts."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "AWie2TQnaaEU",
    "outputId": "0ebd6e4d-5f8f-42ad-dbd4-4325897904a7"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>domain</th>\n",
       "      <th>domain_distinct</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>000directory</td>\n",
       "      <td>riyect0do</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>000webhost</td>\n",
       "      <td>bshet0ow</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>001fans</td>\n",
       "      <td>1sfn0a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>01-telecharger</td>\n",
       "      <td>1rghect0l-a</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>010shangpu</td>\n",
       "      <td>1sghnu0ap</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           domain domain_distinct\n",
       "0    000directory       riyect0do\n",
       "1      000webhost        bshet0ow\n",
       "2         001fans          1sfn0a\n",
       "3  01-telecharger     1rghect0l-a\n",
       "4      010shangpu       1sghnu0ap"
      ]
     },
     "execution_count": 86,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[\"domain_distinct\"] = data_df[\"domain\"].apply(lambda x: ''.join(set(x)))\n",
    "data_df[[\"domain\", \"domain_distinct\"]].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "uoONAJy5dtUe"
   },
   "outputs": [],
   "source": [
    "data_df[\"domain_distinct_vowel_count\"] = data_df[\"domain_distinct\"].apply(lambda x: len([w for w in x if w in vowels]))\n",
    "data_df[\"domain_distinct_vowel_ratio\"] = data_df[\"domain_distinct_vowel_count\"] / data_df[\"domain_distinct\"].apply(len)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ZwiSYXKngI-F"
   },
   "source": [
    "Lets look at the overlap again"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 333
    },
    "colab_type": "code",
    "id": "pS_sCH_odt_y",
    "outputId": "0302deb0-e4f9-4f98-bb18-8fd6d518f8ef"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "class\n",
       "dga      AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "legit    AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "Name: domain_distinct_vowel_ratio, dtype: object"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_df[[\"class\", \"domain_distinct_vowel_ratio\"]].groupby(\"class\")[\"domain_distinct_vowel_ratio\"].plot.kde(legend = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 142
    },
    "colab_type": "code",
    "id": "Ni_AY0lha1uK",
    "outputId": "0e45674a-4e66-4a0f-965d-d91da35c6920"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>class</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>dga</th>\n",
       "      <td>52665.0</td>\n",
       "      <td>0.221543</td>\n",
       "      <td>0.096206</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.166667</td>\n",
       "      <td>0.214286</td>\n",
       "      <td>0.277778</td>\n",
       "      <td>0.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>legit</th>\n",
       "      <td>81261.0</td>\n",
       "      <td>0.358638</td>\n",
       "      <td>0.097000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.285714</td>\n",
       "      <td>0.357143</td>\n",
       "      <td>0.428571</td>\n",
       "      <td>0.833333</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         count      mean       std  min       25%       50%       75%  \\\n",
       "class                                                                   \n",
       "dga    52665.0  0.221543  0.096206  0.0  0.166667  0.214286  0.277778   \n",
       "legit  81261.0  0.358638  0.097000  0.0  0.285714  0.357143  0.428571   \n",
       "\n",
       "            max  \n",
       "class            \n",
       "dga    0.666667  \n",
       "legit  0.833333  "
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domain_distinct_vowel_ratio_stats_df = data_df[[\"class\", \"domain_distinct_vowel_ratio\"]].groupby(\"class\")[\"domain_distinct_vowel_ratio\"].describe()\n",
    "domain_distinct_vowel_ratio_stats_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "G-5RrZhbgPDB",
    "outputId": "95704f13-6fce-4252-965b-d9fdd970a919"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.1596829967635151"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "number_of_dga_domain_distinct_vowel_ratio_overlap = len(data_df[(data_df[\"class\"] == \"dga\") & (data_df[\"domain_distinct_vowel_ratio\"] >= domain_by_class_vowel_ratio_stats_df.loc[\"dga\"][\"75%\"])])\n",
    "number_of_dga_domain_distinct_vowel_ratio_overlap / len(data_df)\n",
    "number_of_dga_domain_distinct_vowel_ratio_overlap / len(data_df[(data_df[\"class\"] != \"dga\") & (data_df[\"domain_vowel_ratio\"] >= 0)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "bOXmT2W3gvYL"
   },
   "source": [
    "It would seem we've reduced the overlap by ~ 2 percent for `dga` records having a `domain_distinct_vowel_ratio` above the 75th percentile overlapping with `legit` records."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 308
    },
    "colab_type": "code",
    "id": "ijMsoHPqbUUU",
    "outputId": "b8c44e65-31f0-4f73-a643-0a17aa7c3bfe"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x209060cab38>"
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "domain_distinct_vowel_ratio_stats_df[domain_distinct_vowel_ratio_stats_df.columns[1:]].plot.bar()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "VI9KUXQfiXbC"
   },
   "source": [
    "### Observations on Vowel Ratios\n",
    "\n",
    "* Vowel Ratios have some discriminatory value between the `class` values.\n",
    "* Vowlel Ratios that are calculated on the distinct character counts provide a nominal improvement."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "fUKOfwfaiwKn"
   },
   "source": [
    "## Digits\n",
    "\n",
    "As we saw during Vowel Ratio analysis domains with a low to zero vowel count tended to have numbers present.\n",
    "\n",
    "Let's generate similar measures for the counts of digits present."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "5RIjlD1bjRj4"
   },
   "outputs": [],
   "source": [
    "digits = [\"0\",\"1\",\"2\",\"3\",\"4\",\"5\",\"6\",\"7\",\"8\",\"9\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "nJXfJGQtjCJZ"
   },
   "outputs": [],
   "source": [
    "data_df[\"digit_count\"] = domain_letter_split.apply(lambda x: len([w for w in x if w in digits]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 170
    },
    "colab_type": "code",
    "id": "ZjluDkQ-jZgP",
    "outputId": "83d2079d-f505-44ec-a16a-68bcdb1e7f88"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    133926.000000\n",
       "mean          0.759136\n",
       "std           2.466659\n",
       "min           0.000000\n",
       "25%           0.000000\n",
       "50%           0.000000\n",
       "75%           0.000000\n",
       "max          21.000000\n",
       "Name: digit_count, dtype: float64"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[\"digit_count\"].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 333
    },
    "colab_type": "code",
    "id": "W8J93SyejfhZ",
    "outputId": "0799f917-5355-4844-a003-8cc133a29d24"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "class\n",
       "dga      AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "legit    AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "Name: digit_count, dtype: object"
      ]
     },
     "execution_count": 95,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_df[[\"class\", \"digit_count\"]].groupby(\"class\")[\"digit_count\"].plot.kde(legend = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Y84qRvVPkBYh"
   },
   "source": [
    "That's a bit difficult to read, let's limit it to just the records where `domain_vowel_ratio` is zero."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 333
    },
    "colab_type": "code",
    "id": "4JnOCbqljucx",
    "outputId": "b74d2eba-2746-4353-b09e-361b5accde9a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "class\n",
       "dga      AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "legit    AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "Name: digit_count, dtype: object"
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAD4CAYAAADhNOGaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nO3deXxU9bn48c8zWyZ7AknYwqYgoFQF44JbtVVRq1Krvbj0ttpr0aq37e2t99ra22o39daq99V6S631Vq0W/bkVFevS2qLFBXBBEBFUhICQQPZlkkzm+f1xZmII2TNnJjPzvF+vec3MOWfOeRjjeea7i6pijDEmc3mSHYAxxpjkskRgjDEZzhKBMcZkOEsExhiT4SwRGGNMhvMlO4ChKikp0WnTpiU7DGOMSSlr167do6qlve1LuUQwbdo01qxZk+wwjDEmpYjIR33ts6ohY4zJcJYIjDEmw1kiMMaYDJdybQTGGDNcHR0dVFZWEgqFkh2Ka4LBIOXl5fj9/kF/xhKBMSZjVFZWkp+fz7Rp0xCRZIcTd6rK3r17qaysZPr06YP+nFUNGWMyRigUYuzYsWmZBABEhLFjxw65xGOJwBiTUdI1CcQM599nicD0am9TG/es2kp1Y1uyQzHGuMwSgenVNQ+v44fLN/BvD76Z7FCMSVvXX389t9xyS7LDsERg9re3qY2/baoC4KUte6hqSN8eFsYYSwSmF399t4qIws/PPxSAlZv3JDkiY9LHT3/6U2bNmsUpp5zCpk2bAFi9ejWHHnooCxYs4JprrmHu3LkAbN26lRNOOIH58+czf/58Vq1a5UpM1n3U7OftHfXkZfn4wvxyfrZiI699uJfzjyhPdljGxNUNT2zgnZ0NcT3nwRML+OHZh/S5f+3atSxbtow33niDcDjM/PnzOeKII7j00ku58847OfbYY7n22mu7ji8rK+O5554jGAyyefNmLrzwQlfmWrNEYPazfkc9B08swOsR5kwoYNOuxmSHZExaePHFFzn33HPJyckB4JxzzgGgsbGRY489FoCLLrqIJ598EnAGwF199dW8+eabeL1e3nvvPVfiskRg9hGJKBs/buSCoyYDMGt8Pste204kong86d3tzmSW/n65u6ln905V7fPY2267jXHjxvHWW28RiUQIBoOuxGRtBGYfVY1ttHZ0cmBpHgCzx+fT2tHJtpqWJEdmTOo78cQTeeyxx2htbaWxsZEnnngCESE/P59XXnkFgGXLlnUdX19fz4QJE/B4PNx33310dna6EpclArOP2A1/8hin6Dq9xEkIH1kiMGbE5s+fz+LFizn88MM577zzOOGEEwD43e9+x5IlS1iwYAGqSmFhIQBXXnkl99xzD8cccwzvvfceubm5rsRlVUNmH9tjiaA423kek73PdmPMyFx33XVcd911+2xrampi3bp1ANx0001UVFQAMHPmzK7tADfeeKMrMblaIhCR00Vkk4hsEZFr+znuSBHpFJHz3YzHDGxbTQsiMCmaCMblBwl4PZYIjHHRU089xeGHH87cuXN58cUX+f73v5/Q67tWIhARL3AHcCpQCawWkeWq+k4vx90MPONWLGbwtte2ML4gSJbPC4DHI5QXZ7O91hKBMW5ZvHgxixcvTtr13SwRHAVsUdUPVLUdWAYs6uW4fwUeAapcjMUM0vaalq72gZhJxdlU1rYmKSJjjNvcTASTgO3d3ldGt3URkUnAucDS/k4kIktEZI2IrKmuro57oOYTlbWtlEerhWLGFwTZbdNMGJO23EwEvXU679lh9nbgP1W13z5RqnqnqlaoakVpaWncAjT76owoVY1tTCjct6/yuIIge5ra6Yz03d/ZGJO63Ow1VAlM7va+HNjZ45gKYFl0gEUJcKaIhFX1cRfjMn2oaXZu9mX5PRNBFp0RZW9z2377jDGpz80SwWpgpohMF5EAcAGwvPsBqjpdVaep6jTgYeBKSwLJE1t7oCw/a5/tZQXOzb+qwdYmMGak8vLyhv3ZH/zgBzz//PMA3H777bS0xKcTh2uJQFXDwNU4vYE2Ag+p6gYRuUJErnDrumb4qhqddoCygn0TwbhoIrB2AmOS60c/+hGnnHIKEN9E4OqAMlVdAazosa3XhmFVvcTNWMzAqqIlgtK8/auGuu83xsTHz3/+cx566CHa2to499xzueGGGwD48Y9/zP3338/kyZMpKSnhiCOO4Dvf+Q6XXHIJZ511Fjt37mTnzp2cfPLJlJSU8MILL4woDhtZbLp0VQ31KBGU5GUhYiUCk2aevhZ2vR3fc47/FJxx06AOffbZZ9m8eTOvvfYaqso555zDypUrycnJ4ZFHHtlvquruvvGNb3DrrbfywgsvUFJSMuKwLRGYLtWNbeQHfQT93n22+70exuYG2G1tBMbEzbPPPsuzzz7LvHnzAGeaic2bN9PY2MiiRYvIzna6cZ999tmux2KJwHSpagzt11AcU5ofpLrRSgQmjQzyl7tbVJXvfve7XH755ftsv+222xIei80+arpUNfTdPXRsboCa5vYER2RM+lq4cCF33303TU1NAOzYsYOqqiqOP/54nnjiCUKhEE1NTTz11FO9fj4/P5/GxvgsGmUlAtNlb3M7cycV9rpvTG6ASptvyJi4Oe2009i4cSMLFiwAnG6lf/jDHzjyyCM555xzOOyww5g6dSoVFRVd01J3t2TJEs444wwmTJgw4sZi6W91nNGooqJC3Viz08BhNzzL5w+fyA2L5u637/rlG3j09UrWXb8wCZEZEx8bN25kzpw5yQ5jQE1NTeTl5dHS0sKJJ57InXfeyfz58wf9+d7+nSKyVlUrejveSgQGgHBnhPrWDopzA73uL84J0BAK09EZwe+1GkVj3LRkyRLeeecdQqEQX/nKV4aUBIbDEoEBoK61A3CqgHozJs/ZXtvSbtNMGOOyBx54IKHXs592BoDaaENwcU4fiSC6vba5I2ExGeOGVKsOH6rh/PssERiArh5BfZUIinP9AOxttrEEJnUFg0H27t2btslAVdm7dy/B4NBK7VY1ZACnygf6LhGMzXXGF1iJwKSy8vJyKisrSed1TYLBIOXl5UP6jCUCA0BNc/9tBLESQU2LjSUwqcvv9zN9+vRkhzHqWNWQAT4pERTl+HvdHysp1DRZIjAm3VgiMIDTRpAb8O43z1CM3+uhIOjrShjGmPRhicAATomgqI/2gZgxuQH22jQTxqQdSwQGcLqP9tU+EFOcG+jqZmqMSR+WCAwANS19jyqOsYnnjElPlggMEC0R9NFQHFOcY4nAmHRkicAATiIYqERQlOOnrtUSgTHpxhKBoT0cobEt3OdgspjCbD+hjght4c4ERWaMSQRLBIb66IRzxQNUDRVm+/c53hiTHiwRmK4be0F2/4kgtr/BEoExacUSgaEhNLhEYCUCY9KTJQLzSYkg2H8iiA04s0RgTHqxRGC6qnoKrURgTEayRGCGnAjqWiwRGJNOLBGYbo3F/c9KXhD07XO8MSY9WCIwNITCBP0esny9zzwa4/N6yMvyWSIwJs1YIjDUt3QMWC0UU5jtt0RgTJqxRGCob+0YsMdQTGG238YRGJNmLBEYGkJWIjAmk1kiME6JYAiJwHoNGZNeLBEY6lutRGBMJrNEYGgYSiLIsURgTLqxRJDhIhGlsS3cNUZgIIXZftrCEUIdNhW1MenCEkGGawyFUR14wrmYQpuB1Ji0Y4kgw8VmHh1KGwHY6GJj0oklggw32LUIYrrmG7JEYEzacDURiMjpIrJJRLaIyLW97F8kIutE5E0RWSMix7sZj9lf/SAnnIvpKhFYF1Jj0sbgWgiHQUS8wB3AqUAlsFpElqvqO90O+wuwXFVVRA4FHgJmuxWT2V/DINciiLGqIWPSj5slgqOALar6gaq2A8uARd0PUNUmVdXo21xAMQnVVSIYYL3iGEsExqQfNxPBJGB7t/eV0W37EJFzReRd4Cngq72dSESWRKuO1lRXV7sSbKYaatVQfrSbaayR2RiT+txMBNLLtv1+8avqY6o6G/g88OPeTqSqd6pqhapWlJaWxjnMzNYQ6sDrEXID/U9BHWNTURuTftxMBJXA5G7vy4GdfR2sqiuBA0WkxMWYTA/OzKM+RHrL271zZiANuxiVMSaR3EwEq4GZIjJdRALABcDy7geIyAyJ3oFEZD4QAPa6GJPpob41POhqoZj8oM+qhoxJI671GlLVsIhcDTwDeIG7VXWDiFwR3b8UOA/4soh0AK3A4m6NxyYBGoYw82iMTTxnTHpxLREAqOoKYEWPbUu7vb4ZuNnNGEz/hjLzaExBtp/tNS0uRWSMSTQbWZzhGoawOllMQdBWKTMmnVgiyHANoeFVDTWErLHYmHRhiSCDqeowq4Z8NLWFCXdGXIrMGJNIlggyWKgjQkenUpA9tKaiWOJotFKBMWnBEkEGG+qo4phYm4J1ITUmPVgiyGD1Q5xwLsbmGzImvVgiyGBDXZQmpqBrlTKrGjImHVgiyGANQ1yUJibWpmBVQ8akB0sEGSx2Ix/swvUxVjVkTHqxRJDBYlU7Qy4RBG0Be2PSiSWCDBa7kecPsUSQE/Di84iVCIxJE5YIMlhDqIOg30OWb3BrEcSICAXZfmsjMCZNWCLIYA2t4SF3HY0pCPqs15AxacISQQYbzjxDMTYVtTHpwxJBBmsIDX2eoRirGjImfVgiyGBO1dDwlqQosBKBMWnDEkEGG0nVkLMmgbURGJMOLBFksOEsShNTkG3rFhuTLgaVCETkERH5nIhY4kgTqkpDKDzkKahjCrP9tIcjhDo64xyZMSbRBntj/zVwEbBZRG4SkdkuxmQSoKW9k86IjqD7qI0uNiZdDCoRqOrzqnoxMB/YCjwnIqtE5FIRGd6dxCRV/TAnnIux+YaMSR+DruoRkbHAJcBlwBvA/+Akhudcicy46pMJ54bffbT7eYwxqWtQFcQi8igwG7gPOFtVP47uelBE1rgVnHHPJxPODbP7aLTbqfUcMib1DfYucJeqrui+QUSyVLVNVStciMu4rGGYq5PFWNWQMeljsFVDP+ll28vxDMQkVlfV0AhGFnc/jzEmdfVbIhCR8cAkIFtE5gES3VUA5Lgcm3HRJyWC4VYNRUsELZYIjEl1A90FFuI0EJcDt3bb3gh8z6WYTAI0hJy6/fxhVg0FfB6y/V4rERiTBvpNBKp6D3CPiJynqo8kKCaTAA2tHWT7vQR8wx8jWJBtU1Ebkw4Gqhr6kqr+AZgmIt/uuV9Vb+3lYyYFOPMMDa9aKMamojYmPQx0J8iNPue5HYhJrIbW8LCnoI4pCNpU1Makg4Gqhn4Tfb4hMeGYRGkIDX/CuZjCbD+7GkJxisgYkyyDnXTuv0WkQET8IvIXEdkjIl9yOzjjnpFMQR1ji9MYkx4G21J4mqo2AGcBlcBBwDWuRWVcN5JFaWJs3WJj0sNgE0Hsp+OZwB9VtcaleEyCxKNEUBgtEUQiGqeojDHJMNhE8ISIvAtUAH8RkVLAKodTlKqOaFGamIJsP6rQ1G6lAmNS2WCnob4WWABUqGoH0AwscjMw456mtjARHf6EczGxEoWNLjYmtQ3lTjAHZzxB98/cG+d4TALERhWPuEQQtPmGjEkHg+01dB9wC3A8cGT0MeCsoyJyuohsEpEtInJtL/svFpF10ccqETlsiPGbYWgY4aI0MbEShTUYG5PaBlsiqAAOVtVBtwqKiBe4AzgVp6fRahFZrqrvdDvsQ+DTqlorImcAdwJHD/YaZnhGOgV1jE1FbUx6GGxj8Xpg/BDPfRSwRVU/UNV2YBk92hVUdZWq1kbfvoIzuZ1xWVfV0EjbCKxqyJi0MNg7QQnwjoi8BrTFNqrqOf18ZhKwvdv7Svr/tf8vwNO97RCRJcASgClTpgwyZNOXuJUIcmwBe2PSwWATwfXDOLf0sq3XqiURORknERzf235VvROn2oiKigrrtD5CI12UJiYv4EPEEoExqW5QiUBV/y4iU4GZqvq8iOQA3gE+VglM7va+HNjZ8yARORS4CzhDVfcOLmwzErHG3fwRjiz2eIT8LF9XVZMxJjUNttfQ14CHgd9EN00CHh/gY6uBmSIyXUQCwAXA8h7nnQI8Cvyzqr43lMDN8DWEOsgJePF7h78WQUxhjk1FbUyqG+xPwqtwGn9fBVDVzSJS1t8HVDUsIlcDz+CUHu5W1Q0ickV0/1LgB8BY4H9FBCCsqgN2SzUjUx+HUcUxBUG/VQ0Zk+IGmwjaVLU9erMmOqhswLp6VV0BrOixbWm315cBlw06WhMX9a0dFOXEJxHY4jTGpL7B1g38XUS+h7OI/anA/wOecC8s46b6lo4RL0oTY4vTGJP6BpsIrgWqgbeBy3F+5X/fraCMu+pa2+NWIijI9lmJwJgUN9heQxEReRx4XFWrXY7JuKyupYOi7EBczlWY7bcpJoxJcf2WCMRxvYjsAd4FNolItYj8IDHhmXhTVeri2EZQEPTT2tFJezgSl/MZYxJvoKqhbwHHAUeq6lhVHYMzOvg4Efk316MzcRfqiNAejnSNCh6prtHF1k5gTMoaKBF8GbhQVT+MbVDVD4AvRfeZFBOrz49X1VDXfEPWTmBMyhooEfhVdU/PjdF2gvj8pDQJVdfaDhDXxmKwGUiNSWUDJYL2Ye4zo1RdS6xEEL9xBIBNM2FMChuo19BhItLQy3YBgi7EY1wWSwTxaiOwqiFjUl+/iUBVB5pYzqSY+q6qofh1H3XOa4nAmFQ18lnHTEqJd9VQQbb1GjIm1VkiyDB1rR34vUJOID6FvSyfh4DXYyUCY1KYJYIMU9/aQWF2gNgEgiMlIhTY6GJjUpolggxT3xK/UcUxBdk+ayw2JoVZIsgwda3tcWsfiCnK9neNTzDGpB5LBBmmzoUSwZjcALXNViIwJlVZIsgwdS0dI160vqfinAC1LVYiMCZVWSLIMPWt8ZuCOmZMboCa5nZUB1y0zhgzClkiyCAdnRGa2sJxrxoqzg3QFo7Q2tEZ1/MaYxLDEkEGifXsiXsbQXSUck2zVQ8Zk4osEWSQWD1+vNYrjinOdRKBNRgbk5osEWSQmuiNemxuVlzPOybXSSw11mBsTEqyRJBBaprbAKdxN56Kc2IlAksExqQiSwQZZG/0Rh3vRBA7n7URGJOaLBFkkNgv9uLcOE8xEfTjEWwsgTEpyhJBBtnb3E5+lo8sX3yXmfB4hOKcgJUIjElRlggySE1ze1cPn3gryvFbicCYFGWJIIPUNLfHvX0gJja62BiTeiwRZJCa5nbGDicRtNZC8x7oZwqJ4hybeM6YVDXQ4vUmjdQ0tzNnQsHgP7DtFXjme7BjrfO+ZBaceA186nzosbDNmNwAb2yvi2O0xphEsRJBhlDVoZUI3rgf/u8MaKqGz/wXnPYT8GXBo5fB41+Hzn1//RfnBqi1ieeMSUlWIsgQLe2dtIUjg2sj2Pw8/OkqOOAkWHwfZOU724+5Elb+HP52I7Q3wRfvAY/TA2lMToBwRGlsC1MQjG/3VGOMu6xEkCFqusYQDJAIGnfBY0ug7GC44IFPkgA4N/2TroWFP4ONT8BzP+ja9cl8Q9ZgbEyqsUSQIWKJYMCqoWeug/Zm+OLvIZDT+zELroKjlsDLv4K3Hwa6zTdkicCYlGOJIEPUDGZ6iY9WwfqH4dhvQOlB/Z9w4Y1QfhQ8+W2o/YiSPGciuz1NlgiMSTWWCDLEoOYZ+suPIX8CHP+tgU/o9cF5vwWNwKNLKM112gqqG9viEa4xJoEsEWSIqsYQAGX5wd4P2PoP2LYKjvsWBHIHd9LiaXDWrbD9FUrf+jVgicCYVORqIhCR00Vkk4hsEZFre9k/W0ReFpE2EfmOm7FkuurGNvKzfGQH+phnaOXPIbcU5n95aCc+9J9g7nn4Vt7MsdnbqG4KjTxYY0xCuZYIRMQL3AGcARwMXCgiB/c4rAb4BnCLW3EYR1VjG6X5fSxIU/UufPACHH1F3w3E/fncLyC3jJvkV9TV148sUGNMwrlZIjgK2KKqH6hqO7AMWNT9AFWtUtXVgM1N4LLqxjZK+koEa+4GbwDmf2V4J88uhs//L1MilZyx6zfDD9IYkxRuJoJJwPZu7yuj20wSVDe2UdZbImhrgrf+CAcvgrzS4V/gwJP5e/F5fK51OWz5y/DPY4xJODcTgfSybVjzD4jIEhFZIyJrqqurRxhWZqruq2po/cPQ1gBHXjbia7xywNVs0Uno41dCS82Iz2eMSQw3E0ElMLnb+3Jg53BOpKp3qmqFqlaUlo7gV2uGamkP09QW7r3H0Np7nFHEk48e8XWKCwv5ZvuV0LIHnvhmv7OVGmNGDzcTwWpgpohMF5EAcAGw3MXrmT7EunTuVyLYsxl2vg6HX7TfbKLDUZqfxQadTs3R18LG5fC3m0Z8TmOM+1ybdE5VwyJyNfAM4AXuVtUNInJFdP9SERkPrAEKgIiIfAs4WFUb3IorE1VFE8F+bQTrHgIE5p4fl+uU5jklji0zLmVs64fw95sgfzxUXBqX8xtj3OHq7KOqugJY0WPb0m6vd+FUGRkX9VoiUIV1D8IBn4aCCXG5Tuz81c3tcNbt0FQFT34L2hrh2H8dXqkj3AYIeP1xKbUYY/Zn01BngKqG2Kjibolg+2tQ9xGc9N24XSeWCKoa2sAXgAvuh0e/Bs/9l7PIzRk3Q9Hk3j+sCrVbYdvLULnaWQynZiu0Rccl+IIwdgaUV8CMU2HGKeDvY5S0MWZILBFkgOqmNrweoTin2zxD65aBLxvmnBW36xTn+MnyedgVTTz4suD838OrS+H5H8Ivn4ODTodpJ0DBRIiEnZv/7g3w0T+gYYfzuawCmDgPDlsMuWVO/7PWOqjeBG8/Amt/D9ljnLaNiq/C2APj9m8wJhNZIsgAu+qdMQQeT7RqJdwO6x91kkD39QZGSESYUBhkZ13rJxs9HlhwJRx8Dqz6JWx4zGlI7i5vPExdAFOPcx6ls53P9SbcDltfhNfvcRLMy3fAQQvhmK/D9E9b9ZExw2CJIAPsrGtlUlH2Jxu2PAehOjh0cdyvNaEwm4/re5lvqLDcqRo6/SZo2u0sgOPxOduziwZ/AV8AZnzWeTTuckZFr/4d3LvI6QZ79BXO/Ef+7IHPZYwBLBFkhJ31rRxW3u1mu+5BZ4K5A06O+7UmFAV5+f29fR8g4vQkyh8/8ovlj4eTvwfHfxvWPwKv/hqe+AY8fz0ccQnMORvGH+pMmd1dJOIko5oPoPZD57nmQ6dx2+uDgkkw7hCnGsuqnUwGsESQ5iIR5eO6EGfMjf5CDjXApj87N8qeN8g4mFSUze6GEOHOCD5vgmY59wdh3sVOm8FHq+CV/4V/3A4v3eq0gxSWQ84YiHRCaw3U74DObtNlixeKpzpVVO0tzhQZb94Pz3wPph7vLM85/YTE/FuMSQJLBGluT3Mb7Z0RJhVFe9i8+5RzE/xUfMYO9DShMJuIOmMXJhYluHpGBKYd5zwadzkN0Dteh/rtTmOzx+v0Wpp9FhRNgTEHwJjpUDjZ6Z7aXd122PAovPJruOcsOPxiZ63moVRjGZMiLBGkuZ11Tn191015/cPOTbD8SFeuNyGacD6ub018IugufzzMPc95DEfRZDjum3DU5bDyv+Gl250usBc9CCUz4xurMUlmK5SluVgPnolF2dC8B95/wRlJ7FLvmomFzs1/R12aLFDjD8JnfwCXroBQPfz2s071kzFpxBJBmtsnEWx4DLTTtWoh6FYi6N6FNB1MOQaWvAB5ZfCH852lPY1JE5YI0tyOulbysnwUBH3w9sNOF8txh7h2vYKgn/wsX+9dSFNd0RS45Cmn8fn+852qImPSgCWCNLezrpWJRUGkvhK2vzL8OvMhmFScTWVti+vXSYr8cXDJk87I6AcWO8t8GpPiLBGkucraaKPt+kecDQlIBFPH5vDhnmbXr5M0eWXwpUecKTT+8AWnO6oxKcwSQRpTVbbuaWba2FynWqj8SKe7pMumleSyvaaVzkgaL0xTPA0uftgZl/GH86C1NtkRGTNslgjSWHVTG83tnRyW9THsfjtu6w4MZPrYXNo7I/vOOZSOJhzqzLC6dwv88SLoSMN2EZMRLBGksa17nHr6I2qfdub1mfuFhFx3WkkuQHpXD8Uc8Gk4dylsWwWPXuaMXjYmxVgiSGNb9zTjI8ykbX+CmQuduu0EmB5NBFv3ZkAiAKc77sIbYeMT8PR/2lrNJuXYyOI09uHeZj7jXYe3pdqZiydByvKzyAl4M6NEELPgSmjc6Uy1XTABTvj3ZEdkzKBZIkhjH1Y388/BlyBYCjNPS9h1RYSpY3PZmkmJAOCUH0HjbvjLjyB/gjMJnjEpwBJBGtu9q5JjO1fDoV/ff1I1lx1QksuGnfUJvWbSeTyw6A5oroY/XQ3egKujuI2JF2sjSFOhjk6OrP8zXjph3pcSfv2Z4/L4qKaFlvZwwq+dVLG1mqce66zX/NayZEdkzIAsEaSpLbvq+WfPc+wtqYCyOQm//pwJBajCpl2NCb920gVy4aKHYPqJ8NgVsPquZEdkTL8sEaSp+nVPMtlTTccRX0vK9Q+eUADAxo8zMBEABHLgwmXOespP/Tv8+XvWtdSMWpYI0tSETffysY6h9Ej3p5TozaSibPKyfGz8uCEp1x8V/NlwwQPOOsqv3AHLLoKWmmRHZcx+LBGko93vcEDDav6SexZeX2IbiWM8HmHOhPzMazDuyeOFM26GM29xlsBceoLNWmpGHUsEaSiy8uc0a5AdM5PbffGw8iLW72ygLWxVIhz1NfiXZ53eW/93ptPF1KakMKOEJYJ0s2czsuEx7u08lTkHTE1qKBXTimkPR1i/I4Orh7qbNB8uXwmHLoYXfwFLj7MFbsyoYIkg3ay8hU5PFneFz2Te5OQutH7E1DEAvP6RzczZJVgA5/4avvQodLbD7890xhw0VSc7MpPBLBGkk51vwroHeaFwEb6CMsqLk7h4PFCan8XUsTm8ttUaSPcz47Nw5Suw4Gp464/wyyPglV9DZ0eyIzMZyBJBulCFZ65Dc8Zwfe0ZnDizFHFpgfqhOG5GCau27LF2gt4EcmHhT+Hrq6D8CPjztbD0eHj/hWRHZjKMJVpUEZ0AAA05SURBVIJ0seFR+Oglth32LXaEApx4UGmyIwLgM7PKaG7vZPWHVj3Up9JZTlXRBQ9ARyvc93lYdjHUbk12ZCZDWCJIB03VsOIamDiPByOfwesRjp9RkuyoADh2xlgCPg/Pb9yd7FBGNxGY/Tm46jX4zPfh/b/Cr46C537orIJmjIts0rlUF4nAE9+AtkYi59zBn36/mxNmllCcG0h2ZADkBHycPKuUJ9ft5LrPzcHvzdzfHh2dEXY3hKhubGNvUzt7mtqij3aa28K0hSOEOjrp6Pw0ZRPncl7t3Rz1j9tpfvUe3phxFfWzFzNhTD4zyvIoCCZnfIhJT5YIUt1Lt8KmFbDwRl5rGc+Ouq1cs3BWsqPaxxePmMwzG3bzwrtVnHbI+GSH45qW9jA7aluprGtlZ10rO2pb2RF93lnXyq6GEL0t45yf5SMv6CPo95Ll8+D3evi4M5fXIlcxzX8KV7bfzfHv/oRN7/yen4YvZmXkMMYXBJk5Lo9DJhZy+OQi5k0pYlxBMPH/aJMWLBGksnUPwV9/4qxFfMzXuevetRTl+DntkHHJjmwfJ80qZXxBkLte/JBTDx43Khqxh6O+tYPtNS1sr2lxbvDdbvY761qpbdm3x4/PI4wvDDKpKJtjDhzLpKJsJhZlU5afRUleFiX5WYzNDRD0e/u56klo5Gs0r3uc6X+9nnsbbmZb8QIeLPoaf6sPcNeLHxCOZpcJhcGupDBvSjGfmlQ4wLmNcVgiSFXrHnJmtpx2PCz6Fe983MjzG3fzzc/OJCcwuv6z+rwevn7Sgfxw+QZe2rKHE2aOjobsniIRZVdDiI/2Ojf7j2qau71uoa7HjT4n4GVSUTaTirM5fHIRk4qznffRbWX5QbyekSc98XjIPfwLMPcsWP1bpvz9Zq6p/SrXzD6Lts9/m/V6AG9ur4s+anl6/S7ASURzJhREE0MR86cUM2VMTsomYuMe0RRbX7WiokLXrFmT7DCSp7MD/nYTvHgLTDsBLlxGpz+X85eu4qO9Lfz13z9NUc7oaB/oLtTRycLbVxJR5elvnkheVvKSVSSiVNa28t7uRt6ramTz7iY2VzWypaqJUEek6zivR5hUlM3UsTlMGeM8po7Nobw4h/LibAqz/cm5qbbUwKtL4ZWl0FYPM06Bo5Y4zx4ve5raeHNbHW9sr+WNbXW8tb2O5nan++6Y3ADzupUaDi0vJN/aGzKCiKxV1Ype91kiSCHbXoUV34Fd65zFZj53G/gC/GzFRu5c+QG3LT6Mc+eVJzvKPq3eWsPi37zMcTNK+O2XK1yvtohElB11zg1/c1WT87y7iS1VTbR2fDKuIVbfftC4fKaX5DJ1bA5Tx+QysSiIbzQ3bofqnbUOXlkKzVVQMAkOvxgO+TyUHez0RAI6I8p7uxt5Y1sdb2yr5Y3tdWypagKcQ6aMyWHWuHxmj89n1vgCZo3PZ9rYnNH9bzdDlrREICKnA/8DeIG7VPWmHvsluv9MoAW4RFVf7++cGZcI2pvh3RXwxr3w4UrIGwef+wXMOZuW9jA/fWoj97+6jS8vmMoN5xwy6ov9D63Zzn88vI6DxuXxndNmcfLssj57EqkqrR2dNLSGaQx10BDqoCEUpqG1g8ZQmJb2MKEOp6dNqCNCKNxJUyjM7oYQuxpC7KoP0Rb+5Bf+uIIsZpbld930DxqXx4yyfAqzU/wXcWcHbHoa1v7e6XaKQtFUZy2EKcfA5KOhcN8fCPWtHbwVrU56d1cD7+5qZOue5q7GbL9XKC/OYfKYHKaMyWbKmBwmFmVTkue0a4zNy6Io248nDlVfJjGSkghExAu8B5wKVAKrgQtV9Z1ux5wJ/CtOIjga+B9VPbq/86ZlIugIQagOWuugaTfUvA9734ftr8HONyDSAYWTCVdcRtXsL/FeTYSX39/LI6/vYE9TG5efeAD/cfrsuNRHJ8LfNlXx/cfXU1nbSrbfy9SxORRm+wlHlFBHJ01tn9zsw711s+nB5xGCfi9Bv5ecgJfxBUHGFzqP6SW5zg2/NJ/CnBS/4Q9G425472l49ynY+hJ0tDjbc8ug5CAomQFjDnB+UOSWQl4ZBIsgkEuIAFtqwrwbrSqrrGllW00L22paqG/df+oLj0Bhtp+cgI+cgJecLB85fi+5WV5yAj6Cfg9ZPm/Xc5bP4/SM8nvI8u2/L8sfOya6rfs+n2fU/8gZ7ZKVCBYA16vqwuj77wKo6o3djvkN8DdV/WP0/SbgJFX9uK/zDjcR/P29an785DvE/r1d/2rd52m//dq1Xz953eMr6/mZReFnuKDzSbx04iHS9fBGH55uzz7CBNh/Xd82/LzLNNZyMP9gHquZTUNo3/rrk2eVccWnD6Bi2pghfx/JFu6M8Nd3q1j1/l4qa1tpCHXg9wpZPi/5QR/5QR8FQT/5QT8F2bHXPgqy/RQEfeQH/eRm+Qj6PFaF0ZfOMOx+26lS3P027NnsPFr7mftJPODPcabLFg+IF8RDRDx0qodOhLAKnSp0qlPtpAoRVSJK9OFsU1UiEN3v/E2f1f6zYf9zYgkh6PcS8HnwRBNDLD/E0kQsYXSljYH2p5DFR07mshMOGNZn+0sEbrbYTQK2d3tfifOrf6BjJgH7JAIRWQIsAZgyZcqwgsnL8jFrXH70hPs87feH0e8fVte+vv4IYULDZOrrZhERDyoeFA8R8aLiPEeIvfageAl58wj58mn1FtDsK6Y2OIkGf5nzPyIwWYTJQHFOgLKCLKaMyeGwyUVJbXAdKZ/Xw2mHjE/rcQVJ5/XBxHnOo7tQAzRXQ1OV07YQqnemtmhvdkoQ7S1OKTTSCdoJGsETieDRCP7o+/2X3ezlB2W3X0wKqMfH+rMXEuropC0coS363PW+++te9rV1ve6krSPinLPPH26x93398EutttGYkrwsV87r5p2kt4Tb89sfzDGo6p3AneCUCIYTzBFTizliavFwPjoMh+HUeBkzCgULnMfYAxN2SYk+8iClf8CkKzfL1JXA5G7vy4GdwzjGGGOMi9xMBKuBmSIyXUQCwAXA8h7HLAe+LI5jgPr+2geMMcbEn2tlNFUNi8jVwDM43UfvVtUNInJFdP9SYAVOj6EtON1HL3UrHmOMMb1ztbJOVVfg3Oy7b1va7bUCV7kZgzHGmP5ZvztjjMlwlgiMMSbDWSIwxpgMZ4nAGGMyXMrNPioi1cBHQAmwJ8nhDJXF7L5Uixcs5kRJtZjjHe9UVe11MZCUSwQxIrKmr3kzRiuL2X2pFi9YzImSajEnMl6rGjLGmAxnicAYYzJcKieCO5MdwDBYzO5LtXjBYk6UVIs5YfGmbBuBMcaY+EjlEoExxpg4sERgjDEZLuUSgYh8UUQ2iEhERCp67PuuiGwRkU0isjBZMfZFRK4XkR0i8mb0cWayY+qLiJwe/R63iMi1yY5nMERkq4i8Hf1uR+XC1iJyt4hUicj6btvGiMhzIrI5+pyoFZQGpY+YR+3fsohMFpEXRGRj9F7xzej2Ufs99xNzQr7nlGsjEJE5QAT4DfAdVV0T3X4w8EfgKGAi8DxwkKr2XFMvaUTkeqBJVW9Jdiz9EREv8B5wKs7iQauBC1X1naQGNgAR2QpUqOqoHTQkIicCTcC9qjo3uu2/gRpVvSmadItV9T+TGWd3fcR8PaP0b1lEJgATVPV1EckH1gKfBy5hlH7P/cT8TyTge065EoGqblTVTb3sWgQsU9U2Vf0QZ42DoxIbXdo4Ctiiqh+oajuwDOf7NSOkqiuBnqvHLwLuib6+B+cGMGr0EfOopaofq+rr0deNwEactdBH7ffcT8wJkXKJoB+TgO3d3leSwC9yCK4WkXXR4vaoKZr2kCrfZU8KPCsia0VkSbKDGYJxsZX5os9lSY5nsEb937KITAPmAa+SIt9zj5ghAd/zqEwEIvK8iKzv5dHfr1LpZVvC670GiP3XwIHA4cDHwC8SHd8gjYrvchiOU9X5wBnAVdEqDeOOUf+3LCJ5wCPAt1S1IdnxDEYvMSfke3Z1hbLhUtVThvGxSmByt/flwM74RDR4g41dRH4LPOlyOMM1Kr7LoVLVndHnKhF5DKeKa2VyoxqU3SIyQVU/jtYVVyU7oIGo6u7Y69H4tywifpwb6v2q+mh086j+nnuLOVHf86gsEQzTcuACEckSkenATOC1JMe0j+gfX8y5wPq+jk2y1cBMEZkuIgHgApzvd9QSkdxoIxsikgucxuj9fntaDnwl+vorwJ+SGMugjOa/ZRER4HfARlW9tduuUfs99xVzor7nVOw1dC7wS6AUqAPeVNWF0X3XAV8FwjhFq6eTFmgvROQ+nCKeAluBy2N1lqNNtJva7YAXuFtVf5rkkPolIgcAj0Xf+oAHRmPMIvJH4CScKYZ3Az8EHgceAqYA24AvquqoaZztI+aTGKV/yyJyPPAi8DZOD0OA7+HUuY/K77mfmC8kAd9zyiUCY4wx8ZVOVUPGGGOGwRKBMcZkOEsExhiT4SwRGGNMhrNEYIwxGc4SgTHGZDhLBMYYk+H+P9OcT+jVjz0pAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_df[(data_df[\"domain_vowel_ratio\"] == 0)][[\"class\", \"digit_count\"]].groupby(\"class\")[\"digit_count\"].plot.kde(legend = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "jd1m8LjHKLTU"
   },
   "source": [
    "Based on the above, further exploration of ratios and distinctness is unlikely to yield a better boundary between our classes."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "qnuqhfVe2RlD"
   },
   "source": [
    "## Observations on Digits\n",
    "\n",
    "There is some discriminating value in counting the number of digits, in that where the `vowel_ratio` is zero non legit domains have an increased likelyhood of having a zero `digit_count`."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "obcFQ5oRnPgJ"
   },
   "source": [
    "## Vowel Distances\n",
    "\n",
    "Another hallmark of words that are sensible to humans is the distance between vowel characters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "H1-Fmzsdn0j4",
    "outputId": "121682ae-c957-4307-88fc-8160658f4710"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>domain</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>92284</th>\n",
       "      <td>prof-klub</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>91705</th>\n",
       "      <td>preguntados</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51166</th>\n",
       "      <td>hanselman</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35432</th>\n",
       "      <td>earnparttimejobs</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25817</th>\n",
       "      <td>clickthecity</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 domain\n",
       "92284         prof-klub\n",
       "91705       preguntados\n",
       "51166         hanselman\n",
       "35432  earnparttimejobs\n",
       "25817      clickthecity"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[data_df[\"class\"] == \"legit\"][[\"domain\"]].sample(5, random_state=6)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "OVg6qg2zo-4M",
    "outputId": "753cc46b-6055-4c82-898a-322a2823d798"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>domain</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>29342</th>\n",
       "      <td>cxccqpkrjieafcd</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1108</th>\n",
       "      <td>15g8f6p19cd8h01gar4ueu39g0c</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>123240</th>\n",
       "      <td>vvhakmoabppjlku</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79389</th>\n",
       "      <td>n0wkjoqhy1r9xby98i1buvwne</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>113764</th>\n",
       "      <td>todairobmamciukjnbheqhce</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                             domain\n",
       "29342               cxccqpkrjieafcd\n",
       "1108    15g8f6p19cd8h01gar4ueu39g0c\n",
       "123240              vvhakmoabppjlku\n",
       "79389     n0wkjoqhy1r9xby98i1buvwne\n",
       "113764     todairobmamciukjnbheqhce"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_df[data_df[\"class\"] == \"dga\"][[\"domain\"]].sample(5, random_state=6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "5tG7J6w2pHU6"
   },
   "source": [
    "Simply eyeballing the data above we can see that in some cases of `dga` records the occurances of vowels is more sparse.\n",
    "\n",
    "If we construct a measure of this sparsity it may be heplful in distinguishing between the classes. Additionally we'll measure the distances as if we are reading left to right, this does mean that non left to right languages will be penalised."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "QCCl3dUTpi7Z"
   },
   "outputs": [],
   "source": [
    "from typing import List, Callable\n",
    "from numbers import Number\n",
    "from scipy.stats import mode\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "\n",
    "def character_distance_left_right(text: str, ords_to_search_for: List[int]) -> np.array:\n",
    "    if len(text) == 0:\n",
    "        raise ValueError(\"Must pass in valid value for `text`\")\n",
    "\n",
    "    if len(ords_to_search_for) == 0:\n",
    "        raise ValueError(\"Must pass in valid value for `ords_to_search_for`\")\n",
    "\n",
    "    if not(all([isinstance(x, Number) for x in ords_to_search_for])):\n",
    "        raise TypeError(\"Invalid data type in `ords_to_search_for` expected Number\")\n",
    "\n",
    "    ord_text = np.array([ord(x) for x in text])\n",
    "\n",
    "    match_indexes = np.argwhere(np.isin(ord_text, ords_to_search_for)).flatten()\n",
    "\n",
    "    if match_indexes.size == 0:\n",
    "        return np.array(len(text))\n",
    "\n",
    "    it = np.nditer(ord_text, flags=['f_index'])\n",
    "    last_vowel_index = 0\n",
    "\n",
    "    result = list()\n",
    "    while not it.finished:\n",
    "        result_matches = match_indexes - it.index\n",
    "        forward_matches = result_matches[result_matches >= 0]\n",
    "\n",
    "        if forward_matches.size > 0:\n",
    "            result.append(forward_matches.min())\n",
    "            last_vowel_index = it.index\n",
    "        else:\n",
    "            result.append(it.index - last_vowel_index)\n",
    "\n",
    "        it.iternext()\n",
    "\n",
    "    return np.array(result)\n",
    "\n",
    "\n",
    "def vowel_distance_left_right_in_column(series: pd.Series, distance_function: Callable) -> pd.Series:\n",
    "    if not isinstance(distance_function, Callable):\n",
    "        raise TypeError(\"`distance_function` must be callable\")\n",
    "\n",
    "    vowel_ords = [ord(x) for x in [\"a\", \"e\", \"i\", \"o\", \"u\", \"y\"]]\n",
    "\n",
    "    column_len = series.astype(str).apply(len)\n",
    "    search_for_ords_distances = series.astype(str).apply(lambda x: character_distance_left_right(x, vowel_ords))\n",
    "    search_for_ords_distances = search_for_ords_distances.apply(distance_function) / column_len\n",
    "    search_for_ords_distances = search_for_ords_distances.clip(upper = 1.0)\n",
    "    search_for_ords_distances = search_for_ords_distances.fillna(1.0)\n",
    "\n",
    "    return search_for_ords_distances\n",
    "\n",
    "\n",
    "def vowel_mode_distance_left_right_in_column(series: pd.Series) -> pd.Series:\n",
    "\n",
    "    def get_mode(x: np.array) -> np.float64:\n",
    "        result = mode(x)\n",
    "\n",
    "        if result[1][0] > 1:\n",
    "            return mode(x)[0][0]\n",
    "        else:\n",
    "            # there was no identifiable mode return np.nan\n",
    "            return np.nan\n",
    "\n",
    "    return vowel_distance_left_right_in_column(series, get_mode)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "Nf6z3lddM1ld",
    "outputId": "58085c86-44f4-4b3a-d29f-e70d7ca38c77"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 0, 2, 1, 0], dtype=int64)"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "character_distance_left_right(\"hello\", [ord(x) for x in [\"e\",\"o\"]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "HzlK3Te9bKys",
    "outputId": "7dcda2b1-3164-40fb-aec7-12a674151062"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 0, 1, 2, 3], dtype=int64)"
      ]
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "character_distance_left_right(\"hello\", [ord(x) for x in [\"e\"]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "CaRdnGiQ5yco",
    "outputId": "010b2fa7-235b-4f35-d60e-e571fa19a75c"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([8, 7, 6, 5, 4, 3, 2, 1, 0], dtype=int64)"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "character_distance_left_right(\"zzzzzzzze\", [ord(x) for x in [\"e\"]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "OQi8coWNm-AD",
    "outputId": "d1014687-a5f6-4c32-b982-76edd726151d"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 2, 3, 4, 5, 6, 7, 8], dtype=int64)"
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "character_distance_left_right(\"ezzzzzzzz\", [ord(x) for x in [\"e\"]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Q_uMmWLdpp7I"
   },
   "outputs": [],
   "source": [
    "def search_for_ords_distances(df, column, search_for, label):\n",
    "    result_df = df[[\"class\"]].copy()\n",
    "    search_for_ords = [ord(x) for x in search_for ]\n",
    "    \n",
    "    col_lengths = df[column].apply(len)\n",
    "\n",
    "    result_df[f\"{column}_{label}_distance_mean_ratio\"] = vowel_distance_left_right_in_column(df[column], np.mean)\n",
    "    result_df[f\"{column}_{label}_distance_std_ratio\"] = vowel_distance_left_right_in_column(df[column], np.std)\n",
    "    result_df[f\"{column}_{label}_distance_max_ratio\"] = vowel_distance_left_right_in_column(df[column], np.max)\n",
    "    result_df[f\"{column}_{label}_distance_min_ratio\"] = vowel_distance_left_right_in_column(df[column], np.min)\n",
    "    result_df[f\"{column}_{label}_distance_median_ratio\"] = vowel_distance_left_right_in_column(df[column], np.median)\n",
    "    result_df[f\"{column}_{label}_distance_mode_ratio\"] = vowel_mode_distance_left_right_in_column(df[column])\n",
    "    \n",
    "    return result_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "dDZLj-ivqS2t"
   },
   "outputs": [],
   "source": [
    "distances_df = search_for_ords_distances(data_df, \"domain\", [\"a\",\"e\", \"i\", \"o\", \"u\"], \"vowel\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 224
    },
    "colab_type": "code",
    "id": "SgMUorq7x-LX",
    "outputId": "e741d74f-72a0-4109-f3d8-177d98f6b652"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>class</th>\n",
       "      <th>domain_vowel_distance_mean_ratio</th>\n",
       "      <th>domain_vowel_distance_std_ratio</th>\n",
       "      <th>domain_vowel_distance_max_ratio</th>\n",
       "      <th>domain_vowel_distance_min_ratio</th>\n",
       "      <th>domain_vowel_distance_median_ratio</th>\n",
       "      <th>domain_vowel_distance_mode_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>legit</td>\n",
       "      <td>0.104167</td>\n",
       "      <td>0.102768</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.083333</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>legit</td>\n",
       "      <td>0.160000</td>\n",
       "      <td>0.120000</td>\n",
       "      <td>0.400000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.150000</td>\n",
       "      <td>0.100000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>legit</td>\n",
       "      <td>0.265306</td>\n",
       "      <td>0.177914</td>\n",
       "      <td>0.571429</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.285714</td>\n",
       "      <td>0.142857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>legit</td>\n",
       "      <td>0.091837</td>\n",
       "      <td>0.082898</td>\n",
       "      <td>0.285714</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.071429</td>\n",
       "      <td>0.071429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>legit</td>\n",
       "      <td>0.210000</td>\n",
       "      <td>0.157797</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   class  domain_vowel_distance_mean_ratio  domain_vowel_distance_std_ratio  \\\n",
       "0  legit                          0.104167                         0.102768   \n",
       "1  legit                          0.160000                         0.120000   \n",
       "2  legit                          0.265306                         0.177914   \n",
       "3  legit                          0.091837                         0.082898   \n",
       "4  legit                          0.210000                         0.157797   \n",
       "\n",
       "   domain_vowel_distance_max_ratio  domain_vowel_distance_min_ratio  \\\n",
       "0                         0.333333                              0.0   \n",
       "1                         0.400000                              0.0   \n",
       "2                         0.571429                              0.0   \n",
       "3                         0.285714                              0.0   \n",
       "4                         0.500000                              0.0   \n",
       "\n",
       "   domain_vowel_distance_median_ratio  domain_vowel_distance_mode_ratio  \n",
       "0                            0.083333                          0.000000  \n",
       "1                            0.150000                          0.100000  \n",
       "2                            0.285714                          0.142857  \n",
       "3                            0.071429                          0.071429  \n",
       "4                            0.200000                          0.000000  "
      ]
     },
     "execution_count": 106,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "distances_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 317
    },
    "colab_type": "code",
    "id": "h8EPr1vZz6cr",
    "outputId": "0efb7d94-6b95-4f55-f30a-d2526f3ee32a"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>domain_vowel_distance_mean_ratio</th>\n",
       "      <th>domain_vowel_distance_std_ratio</th>\n",
       "      <th>domain_vowel_distance_max_ratio</th>\n",
       "      <th>domain_vowel_distance_min_ratio</th>\n",
       "      <th>domain_vowel_distance_median_ratio</th>\n",
       "      <th>domain_vowel_distance_mode_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>133926.000000</td>\n",
       "      <td>133926.000000</td>\n",
       "      <td>133926.000000</td>\n",
       "      <td>133926.000000</td>\n",
       "      <td>133926.000000</td>\n",
       "      <td>133926.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.127635</td>\n",
       "      <td>0.103519</td>\n",
       "      <td>0.324496</td>\n",
       "      <td>0.008647</td>\n",
       "      <td>0.117080</td>\n",
       "      <td>0.041880</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.108137</td>\n",
       "      <td>0.049880</td>\n",
       "      <td>0.172109</td>\n",
       "      <td>0.092584</td>\n",
       "      <td>0.105785</td>\n",
       "      <td>0.120629</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.010000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.045455</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.073964</td>\n",
       "      <td>0.070511</td>\n",
       "      <td>0.200000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.071429</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.102041</td>\n",
       "      <td>0.092807</td>\n",
       "      <td>0.285714</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.100000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.148148</td>\n",
       "      <td>0.125000</td>\n",
       "      <td>0.400000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.142857</td>\n",
       "      <td>0.071429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.288491</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       domain_vowel_distance_mean_ratio  domain_vowel_distance_std_ratio  \\\n",
       "count                     133926.000000                    133926.000000   \n",
       "mean                           0.127635                         0.103519   \n",
       "std                            0.108137                         0.049880   \n",
       "min                            0.010000                         0.000000   \n",
       "25%                            0.073964                         0.070511   \n",
       "50%                            0.102041                         0.092807   \n",
       "75%                            0.148148                         0.125000   \n",
       "max                            1.000000                         0.288491   \n",
       "\n",
       "       domain_vowel_distance_max_ratio  domain_vowel_distance_min_ratio  \\\n",
       "count                    133926.000000                    133926.000000   \n",
       "mean                          0.324496                         0.008647   \n",
       "std                           0.172109                         0.092584   \n",
       "min                           0.045455                         0.000000   \n",
       "25%                           0.200000                         0.000000   \n",
       "50%                           0.285714                         0.000000   \n",
       "75%                           0.400000                         0.000000   \n",
       "max                           1.000000                         1.000000   \n",
       "\n",
       "       domain_vowel_distance_median_ratio  domain_vowel_distance_mode_ratio  \n",
       "count                       133926.000000                     133926.000000  \n",
       "mean                             0.117080                          0.041880  \n",
       "std                              0.105785                          0.120629  \n",
       "min                              0.000000                          0.000000  \n",
       "25%                              0.071429                          0.000000  \n",
       "50%                              0.100000                          0.000000  \n",
       "75%                              0.142857                          0.071429  \n",
       "max                              1.000000                          1.000000  "
      ]
     },
     "execution_count": 107,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "distances_df.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 255
    },
    "colab_type": "code",
    "id": "DBkWqT1Xu6W3",
    "outputId": "b71f57c5-1e7c-477d-8f3b-be5bd3eddb13"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>domain_vowel_distance_mean_ratio</th>\n",
       "      <th>domain_vowel_distance_std_ratio</th>\n",
       "      <th>domain_vowel_distance_max_ratio</th>\n",
       "      <th>domain_vowel_distance_min_ratio</th>\n",
       "      <th>domain_vowel_distance_median_ratio</th>\n",
       "      <th>domain_vowel_distance_mode_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>domain_vowel_distance_mean_ratio</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.474435</td>\n",
       "      <td>0.862796</td>\n",
       "      <td>0.753417</td>\n",
       "      <td>0.961631</td>\n",
       "      <td>0.734087</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain_vowel_distance_std_ratio</th>\n",
       "      <td>0.474435</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.826758</td>\n",
       "      <td>-0.193823</td>\n",
       "      <td>0.354248</td>\n",
       "      <td>0.040495</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain_vowel_distance_max_ratio</th>\n",
       "      <td>0.862796</td>\n",
       "      <td>0.826758</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.366551</td>\n",
       "      <td>0.760031</td>\n",
       "      <td>0.446619</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain_vowel_distance_min_ratio</th>\n",
       "      <td>0.753417</td>\n",
       "      <td>-0.193823</td>\n",
       "      <td>0.366551</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.779481</td>\n",
       "      <td>0.741782</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain_vowel_distance_median_ratio</th>\n",
       "      <td>0.961631</td>\n",
       "      <td>0.354248</td>\n",
       "      <td>0.760031</td>\n",
       "      <td>0.779481</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.769051</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain_vowel_distance_mode_ratio</th>\n",
       "      <td>0.734087</td>\n",
       "      <td>0.040495</td>\n",
       "      <td>0.446619</td>\n",
       "      <td>0.741782</td>\n",
       "      <td>0.769051</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                    domain_vowel_distance_mean_ratio  \\\n",
       "domain_vowel_distance_mean_ratio                            1.000000   \n",
       "domain_vowel_distance_std_ratio                             0.474435   \n",
       "domain_vowel_distance_max_ratio                             0.862796   \n",
       "domain_vowel_distance_min_ratio                             0.753417   \n",
       "domain_vowel_distance_median_ratio                          0.961631   \n",
       "domain_vowel_distance_mode_ratio                            0.734087   \n",
       "\n",
       "                                    domain_vowel_distance_std_ratio  \\\n",
       "domain_vowel_distance_mean_ratio                           0.474435   \n",
       "domain_vowel_distance_std_ratio                            1.000000   \n",
       "domain_vowel_distance_max_ratio                            0.826758   \n",
       "domain_vowel_distance_min_ratio                           -0.193823   \n",
       "domain_vowel_distance_median_ratio                         0.354248   \n",
       "domain_vowel_distance_mode_ratio                           0.040495   \n",
       "\n",
       "                                    domain_vowel_distance_max_ratio  \\\n",
       "domain_vowel_distance_mean_ratio                           0.862796   \n",
       "domain_vowel_distance_std_ratio                            0.826758   \n",
       "domain_vowel_distance_max_ratio                            1.000000   \n",
       "domain_vowel_distance_min_ratio                            0.366551   \n",
       "domain_vowel_distance_median_ratio                         0.760031   \n",
       "domain_vowel_distance_mode_ratio                           0.446619   \n",
       "\n",
       "                                    domain_vowel_distance_min_ratio  \\\n",
       "domain_vowel_distance_mean_ratio                           0.753417   \n",
       "domain_vowel_distance_std_ratio                           -0.193823   \n",
       "domain_vowel_distance_max_ratio                            0.366551   \n",
       "domain_vowel_distance_min_ratio                            1.000000   \n",
       "domain_vowel_distance_median_ratio                         0.779481   \n",
       "domain_vowel_distance_mode_ratio                           0.741782   \n",
       "\n",
       "                                    domain_vowel_distance_median_ratio  \\\n",
       "domain_vowel_distance_mean_ratio                              0.961631   \n",
       "domain_vowel_distance_std_ratio                               0.354248   \n",
       "domain_vowel_distance_max_ratio                               0.760031   \n",
       "domain_vowel_distance_min_ratio                               0.779481   \n",
       "domain_vowel_distance_median_ratio                            1.000000   \n",
       "domain_vowel_distance_mode_ratio                              0.769051   \n",
       "\n",
       "                                    domain_vowel_distance_mode_ratio  \n",
       "domain_vowel_distance_mean_ratio                            0.734087  \n",
       "domain_vowel_distance_std_ratio                             0.040495  \n",
       "domain_vowel_distance_max_ratio                             0.446619  \n",
       "domain_vowel_distance_min_ratio                             0.741782  \n",
       "domain_vowel_distance_median_ratio                          0.769051  \n",
       "domain_vowel_distance_mode_ratio                            1.000000  "
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame(distances_df.corr())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "UlLLVEJXFmue"
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "Q8r821ghIQhz",
    "outputId": "8e38be14-a884-45c4-d7f3-5fed59510d65"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>class</th>\n",
       "      <th>variable</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>294760</th>\n",
       "      <td>legit</td>\n",
       "      <td>domain_vowel_distance_max_ratio</td>\n",
       "      <td>0.095238</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>271775</th>\n",
       "      <td>dga</td>\n",
       "      <td>domain_vowel_distance_max_ratio</td>\n",
       "      <td>0.720000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>270714</th>\n",
       "      <td>dga</td>\n",
       "      <td>domain_vowel_distance_max_ratio</td>\n",
       "      <td>0.360000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>431833</th>\n",
       "      <td>legit</td>\n",
       "      <td>domain_vowel_distance_min_ratio</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>320015</th>\n",
       "      <td>legit</td>\n",
       "      <td>domain_vowel_distance_max_ratio</td>\n",
       "      <td>0.222222</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        class                         variable     value\n",
       "294760  legit  domain_vowel_distance_max_ratio  0.095238\n",
       "271775    dga  domain_vowel_distance_max_ratio  0.720000\n",
       "270714    dga  domain_vowel_distance_max_ratio  0.360000\n",
       "431833  legit  domain_vowel_distance_min_ratio  0.000000\n",
       "320015  legit  domain_vowel_distance_max_ratio  0.222222"
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "melted_distances_df = distances_df.melt(id_vars=\"class\")\n",
    "melted_distances_df.sample(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "1M5YrMbeIl2O",
    "outputId": "8cc7b2ec-6330-4ad9-c2c3-c6771cee7ea0"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x1728 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "unique_distance_vars = melted_distances_df[\"variable\"].unique()\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,4 * len(unique_distance_vars)), ncols = 1, nrows = len(unique_distance_vars), sharey='all', sharex=True)\n",
    "\n",
    "for i, variable in enumerate(unique_distance_vars):\n",
    "    melted_distances_df[melted_distances_df[\"variable\"] == variable].groupby(\"class\")[\"value\"].plot.kde(\n",
    "        legend = True, \n",
    "        title = variable, \n",
    "        ax = ax[i],\n",
    "        stacked = True,\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "QPjPPuR2PGId"
   },
   "source": [
    "Let's remove `domain_vowel_distance_min_ratio` from consideration, as its peaks between the classes center on the same values.\n",
    "\n",
    "While `domain_vowel_distance_mode_ratio` also shares a peak between the classes there is a some distinction between the class after this first peak. Thus we'll keep it in consideration."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Whv7xpgGOf7h"
   },
   "outputs": [],
   "source": [
    "melted_distances_df = melted_distances_df[~melted_distances_df[\"variable\"].isin([\"domain_vowel_distance_min_ratio\"])]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "TajWRq4WeByX"
   },
   "source": [
    "Let's replot given we're removed features from consideration."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "CTT8AoiDOwfI",
    "outputId": "a5cbd8c5-0368-4585-f0df-932236b051df"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x1440 with 5 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "unique_distance_vars = melted_distances_df[\"variable\"].unique()\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10,4 * len(unique_distance_vars)), ncols = 1, nrows = len(unique_distance_vars), sharey='all', sharex=True)\n",
    "\n",
    "for i, variable in enumerate(unique_distance_vars):\n",
    "    melted_distances_df[melted_distances_df[\"variable\"] == variable].groupby(\"class\")[\"value\"].plot.kde(\n",
    "        legend = True, \n",
    "        title = variable, \n",
    "        ax = ax[i],\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "tr7darhCeHzs"
   },
   "source": [
    "Now a question must be asked of the remaing vowel distance features, which features are highly correlated. That is if two features are highly correlated then we should not include them both as the information gain would not be worth the additional complexity in our model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 224
    },
    "colab_type": "code",
    "id": "mBcz0KHJSC-i",
    "outputId": "82c89ac9-d1d8-4a45-e8e6-fd99557cc582"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>domain_vowel_distance_mean_ratio</th>\n",
       "      <th>domain_vowel_distance_std_ratio</th>\n",
       "      <th>domain_vowel_distance_max_ratio</th>\n",
       "      <th>domain_vowel_distance_median_ratio</th>\n",
       "      <th>domain_vowel_distance_mode_ratio</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>domain_vowel_distance_mean_ratio</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.474435</td>\n",
       "      <td>0.862796</td>\n",
       "      <td>0.961631</td>\n",
       "      <td>0.734087</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain_vowel_distance_std_ratio</th>\n",
       "      <td>0.474435</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.826758</td>\n",
       "      <td>0.354248</td>\n",
       "      <td>0.040495</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain_vowel_distance_max_ratio</th>\n",
       "      <td>0.862796</td>\n",
       "      <td>0.826758</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.760031</td>\n",
       "      <td>0.446619</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain_vowel_distance_median_ratio</th>\n",
       "      <td>0.961631</td>\n",
       "      <td>0.354248</td>\n",
       "      <td>0.760031</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.769051</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>domain_vowel_distance_mode_ratio</th>\n",
       "      <td>0.734087</td>\n",
       "      <td>0.040495</td>\n",
       "      <td>0.446619</td>\n",
       "      <td>0.769051</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                    domain_vowel_distance_mean_ratio  \\\n",
       "domain_vowel_distance_mean_ratio                            1.000000   \n",
       "domain_vowel_distance_std_ratio                             0.474435   \n",
       "domain_vowel_distance_max_ratio                             0.862796   \n",
       "domain_vowel_distance_median_ratio                          0.961631   \n",
       "domain_vowel_distance_mode_ratio                            0.734087   \n",
       "\n",
       "                                    domain_vowel_distance_std_ratio  \\\n",
       "domain_vowel_distance_mean_ratio                           0.474435   \n",
       "domain_vowel_distance_std_ratio                            1.000000   \n",
       "domain_vowel_distance_max_ratio                            0.826758   \n",
       "domain_vowel_distance_median_ratio                         0.354248   \n",
       "domain_vowel_distance_mode_ratio                           0.040495   \n",
       "\n",
       "                                    domain_vowel_distance_max_ratio  \\\n",
       "domain_vowel_distance_mean_ratio                           0.862796   \n",
       "domain_vowel_distance_std_ratio                            0.826758   \n",
       "domain_vowel_distance_max_ratio                            1.000000   \n",
       "domain_vowel_distance_median_ratio                         0.760031   \n",
       "domain_vowel_distance_mode_ratio                           0.446619   \n",
       "\n",
       "                                    domain_vowel_distance_median_ratio  \\\n",
       "domain_vowel_distance_mean_ratio                              0.961631   \n",
       "domain_vowel_distance_std_ratio                               0.354248   \n",
       "domain_vowel_distance_max_ratio                               0.760031   \n",
       "domain_vowel_distance_median_ratio                            1.000000   \n",
       "domain_vowel_distance_mode_ratio                              0.769051   \n",
       "\n",
       "                                    domain_vowel_distance_mode_ratio  \n",
       "domain_vowel_distance_mean_ratio                            0.734087  \n",
       "domain_vowel_distance_std_ratio                             0.040495  \n",
       "domain_vowel_distance_max_ratio                             0.446619  \n",
       "domain_vowel_distance_median_ratio                          0.769051  \n",
       "domain_vowel_distance_mode_ratio                            1.000000  "
      ]
     },
     "execution_count": 114,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "distances_df[unique_distance_vars].corr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 282
    },
    "colab_type": "code",
    "id": "lhJiNClWRveO",
    "outputId": "9cdf2ddd-7ca2-4bcd-c0ec-2055c6273961"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x20913519128>"
      ]
     },
     "execution_count": 115,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "distances_df[unique_distance_vars].corr().abs().sum().sort_values(ascending = False).plot.barh()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "77GUiJY1PzHX"
   },
   "source": [
    "## Observations on Vowel Distances\n",
    "\n",
    "Looking at the correlation table and barplot of absolute correlation magnitudes, we will take `domain_vowel_distance_std_ratio` and `domain_vowel_distance_mode_ratio` forward.\n",
    "\n",
    "Furthermore given the above are expressed as ratios they will not be prone to the issues of length based measures for misclassification."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ToIcfJ-DnVc1"
   },
   "source": [
    "## Most Frequent Consonants\n",
    "\n",
    "We've been paying a fair bit of attention to vowels, however it would also be prudent to consider the consonants.\n",
    "\n",
    "Eyeballing the data `dga` observations a variety of consonants and repeating consonants are not uncommon. Is there anything to this hunch?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Pz8hSZ-gvIpr"
   },
   "outputs": [],
   "source": [
    "from collections import Counter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "UzgaDssaZczp"
   },
   "outputs": [],
   "source": [
    "def get_consonant_variety_ratio(s):\n",
    "    s = \"\".join([c for c in list(s) if c not in [\"a\",\"e\",\"i\",\"o\",\"u\",\"y\"]])\n",
    "  \n",
    "    if len(s) == 0:\n",
    "        return 0.0\n",
    "\n",
    "    counter = Counter(s)\n",
    "    counts = np.array(list(counter.values()))\n",
    "  \n",
    "    if np.min(counts) == np.max(counts):\n",
    "        return 1\n",
    "\n",
    "    return np.max(counts) / np.sum(counts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ESI3FWPkvn-v"
   },
   "outputs": [],
   "source": [
    "data_df[\"domain_consonant_variety_ratio\"] = data_df[\"domain\"].apply(lambda x: get_consonant_variety_ratio(x) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 337
    },
    "colab_type": "code",
    "id": "H_SyNK4GaOKY",
    "outputId": "ee7bb2d6-c33e-4ffd-b6d4-b01873cbe5b1"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "class\n",
       "dga      AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "legit    AxesSubplot(0.125,0.125;0.775x0.755)\n",
       "Name: domain_consonant_variety_ratio, dtype: object"
      ]
     },
     "execution_count": 119,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD8CAYAAAB5Pm/hAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nO3deXxc5XXw8d+jZTTa98WWLO/YGGNjWQbMlkDYE0xJSIGQFKckhk9Dk7R90xeajYamSdO+NEm3lACtExIgYQtr2ELAYANeMV6wLS+yZcvWvo5GGs087x93rjS2RrPeOzPWnO/no8+MZu7cezweHT0693nOVVprhBBCTF0ZyQ5ACCGEvSTRCyHEFCeJXgghpjhJ9EIIMcVJohdCiClOEr0QQkxxtiV6pdQCpdS2gK8+pdTX7TqeEEKI4FQi5tErpTKBo8B5Wutm2w8ohBBiTKJKN58A9kuSF0KIxMtK0HFuBh4Nt1FFRYWeNWuW/dEIIcQUsXnz5g6tdWWobWwv3SilHMAx4Cyt9Ykgz68B1gDU19cvb26WQb8QQkRKKbVZa90YaptElG6uAbYES/IAWusHtNaNWuvGysqQv5SEEELEIBGJ/hYiKNsIIYSwh62JXimVB1wBPGXncYQQQkzO1pOxWmsXUG7nMYQQwuPx0NLSgtvtTnYotnE6ndTV1ZGdnR31axM160YIIWzT0tJCYWEhs2bNQimV7HAsp7Wms7OTlpYWZs+eHfXrpQWCEOK053a7KS8vn5JJHkApRXl5ecx/sUiiF0JMCVM1yZvi+fdJohdR01rzu21HeXtfR7JDESJl3XvvvfzLv/xLssMApEYvYvDyzuN87bFtALz41YtZNL0oyREJIUKREb2I2mMbj1DkzCIrQ/H01pZkhyNEyvj+97/PggULuPzyy9mzZw8AGzduZMmSJaxcuZJvfOMbLF68GIBDhw5x8cUX09DQQENDA+vXr7ctLkn0Iiojoz7eO9DFDctquWBeBa/tbkt2SEKkhM2bN/PYY4+xdetWnnrqKTZu3AjAF7/4RX72s5+xYcMGMjMzx7avqqri1VdfZcuWLTz++ON89atftS02Kd2IqGxv6WHI4+WCeRXMKHPxDy/spq3fTVWhM9mhCQHA3z+3k13H+izd56LpRXz3urNCbrNu3TpuuOEG8vLyAFi1ahUA/f39XHDBBQB87nOf4/nnnweMuf933XUX27ZtIzMzk71791oacyBJ9CIqu1qNH6CldSUU5xoLN3Ye66NqgSR6IU6dGROqaeS//uu/Ul1dzQcffIDP58PptO9nSBK9iMpHx/spzs2muiiHvBzjz9CdR3u5dEFVkiMTwhBu5G2XSy65hNWrV3P33XczOjrKc889xx133EFhYSHvvvsu559/Po899tjY9r29vdTV1ZGRkcHatWvxer22xSY1ehGVPcf7WVBTiFKKImc2M8py+eh4f7LDEiLpGhoauOmmmzjnnHP4zGc+w8UXXwzAQw89xJo1a1i5ciVaa4qLiwH4i7/4C9auXcv555/P3r17yc/Pty02GdGLqBzsGOTqxTVj388qz+dwlyuJEQmROr75zW/yzW9+86THBgYG2L59OwA//OEPaWw0WsfPnz9/7HGAH/zgB7bFJSN6EbHB4VG6BkeYUZo39tis8nwOdgyGrEUKkc5eeOEFzjnnHBYvXsy6dev41re+lfAYZEQvItbSPQTAjLLcscdmlufR7x6lx+WhNN+RrNCESFk33XQTN910U1JjkBG9iNgRf4mmLmBEP7vCqCse7BxMSkxCiPAk0YuItXSbiT5wRG8k+mZJ9EKkLEn0ImJHuofIzc6kPKBEM6MsF6XgUIeckBUiVUmiFxFr6XZRV5p70qKQnKxMKgtyaO0dSmJkQohQJNGLiB3rcVMbULYxTSvJpbV36l7CTYhIFBQUxPza73znO7z22msA/PjHP8blsvYvZEn0ImIn+txUB+lpM63IKYleiDh873vf4/LLLwck0Ysk8vo0HQPDVBXlTHhuWomT1p4hmUsvhN8///M/s2LFCpYsWcJ3v/vdscfvu+8+Fi5cyBVXXMEtt9wydmGS1atX88QTT/DTn/6UY8eOcemll3LppZdaFo+t8+iVUiXAg8BiQAN/rrXeYOcxhT06B4fxaagqnJjopxfnMjjipX94lCJn9FeoF2IqeeWVV9i3bx/vv/8+WmtWrVrFW2+9RV5eHk8++SRbt25ldHSUhoYGli9fftJrv/rVr3L//ffzxhtvUFFRYVlMdi+Y+gnwe631jUopB5AX7gUiNbX1DQNQGaR0U1NsPNba46aoRhK9SLKX7objH1q7z5qz4ZofRrTpK6+8wiuvvMKyZcsAowXCvn376O/v5/rrryc31zjPdd1111kbYwi2lW6UUkXAJcBDAFrrEa11j13HE/Zq6zdq8MFKN9NL/IleZt4Igdaae+65h23btrFt2zaampq4/fbbk1ratHNEPwdoB/5HKbUU2Ax8TWstK2tOQ+aIPljpZlqxMUKRE7IiJUQ48rbLVVddxbe//W1uvfVWCgoKOHr0KNnZ2Vx00UXccccd3HPPPYyOjvLCCy/w5S9/ecLrCwsL6e/vP21KN1lAA/CXWuv3lFI/Ae4Gvh24kVJqDbAGoL6+3sZwRDza+s3SzcREbz5m/jIQIp1deeWV7N69m5UrVwLGtMtHHnmEFStWsGrVKpYuXcrMmTNpbGwca1kcaM2aNVxzzTVMmzaNN954w5KYlF1/TiilaoB3tdaz/N9fDNyttf7kZK9pbGzUmzZtsiUeEZ9vPfMhz29vZdt3rgz6fMN9r3LN4hq+f8PZCY5MCNi9ezdnnnlmssMIa2BggIKCAlwuF5dccgkPPPAADQ0NEb8+2L9TKbVZa90Y6nW2jei11seVUkeUUgu01nuATwC77DqesFdb33DQso2pqjCH9n4Z0QsRypo1a9i1axdut5vbbrstqiQfD7tn3fwl8Cv/jJsDwBdtPp6wSefgCOX5kyf6ysKcsfKOECK4X//610k5rq2JXmu9DQj5J4U4PXQPjnDm9KJJn68syOFAu5xnFyIVycpYERFjRD/5hUUqi3JoHxiW1bEiaab6Zy+ef58kehGWx+ujd8hDaV6IRF+Qw8ioj76h0QRGJoTB6XTS2dk5ZZO91prOzk6czokLFiMhlxIUYXW7RgAoLwiR6P0natsH3BTnyepYkVh1dXW0tLTQ3t6e7FBs43Q6qauri+m1kuhFWN2DHgDKQpRuqvytEdr6hplXVZiQuIQwZWdnM3v27GSHkbKkdCPC6hw0ZtOESvTjI3qZeSNEqpFEL8LqGvSXbsJMrwRZHStEKpJEL8IyE31p/uS19yJnFjlZGTKiFyIFSaIXYY0l+hCzbpRSVMrqWCFSkiR6EVbX4AjFudlkZ4b+uBirY6WDpRCpRhK9CCvcYimT9LsRIjVJohdhdQ2MhJxxY5J+N0KkJkn0Iqxu1wilkST6Aic9Lg8jo74ERCWEiJQkehFWpKWbikKHf3sZ1QuRSiTRi5C01vS6PJSEmHFjqizwL5qS8o0QKUUSvQhpyONlxOujODd8/xpz0VSHzKUXIqVIohch9Q4ZfW5KImhUViEjeiFSkiR6EVKPy5/ooxrRj9gakxAiOpLoRUhmoo+k9bAzO5PCnCwZ0QuRYiTRi5B6h4zReSQ1ekDaIAiRgiTRi5DGa/ThZ90AVBTmSGMzIVKMrRceUUodAvoBLzCqtZYLhZ9moqnRgzHFcndrn50hCSGilIgrTF2qte5IwHGEDXqGPGRlKPIcmRFtX1mYw1v7ZEQvRCqR0o0IqXfIQ0leNkqpiLavKHDQ7x7F7fHaHJkQIlJ2J3oNvKKU2qyUWmPzsYQNel2eiE/EgiyaEiIV2Z3oL9RaNwDXAF9RSl1y6gZKqTVKqU1KqU1T+Qrup6ueoZGIT8RCwLVjZeaNECnD1kSvtT7mv20DngbODbLNA1rrRq11Y2VlpZ3hiBj0DkU3opfVsUKkHtsSvVIqXylVaN4HrgR22HU8YY8elyfiGTcgq2OFSEV2zrqpBp72n8TLAn6ttf69jccTNuh1eSJaFWsqz5cRvRCpxrZEr7U+ACy1a//CfqNeH/3Do1GVbhxZGZTkZcvJWCFSiEyvFJPqc48CkS+WMlUUSBsEIVKJJHoxqR6XUWePZtYNGKtjZUQvROqQRC8m1TMUeefKQJXS70aIlCKJXkzKbGgWTY0epHQjRKqRRC8m1RtlQzNTZWEOrhEvg8OjdoQlhIiSJHoxqVhr9BUFxvZSpxciNUiiF5Mya/RFzuhm4Uq/GyFSiyR6ManeIQ+FOVlkZUb3MZF+N0KkFkn0YlLRroo1VZr9bqQNghApQRK9mFSPvxd9tMryHSglI3ohUoUkejGpaDtXmrIyMyjLc0iiFyJFSKIXk+pxjVCSG92MG1NloayOFSJVSKIXk+odiq1GD/7VsTKiFyIlSKIXQWmtYy7dgLE6Vkb0QqQGSfQiKNeIF49Xx5zozRG91triyIQQ0ZJEL4Iy+9xE2/7AVFHgYHjUx4C0QRAi6STRi6B6XLE1NDPJoikhUockehFUrJ0rTZUFTkASfVrSGkZlsVwqkUQvgjITfVGspZtCs7GZ/MCnFY8b/nMl/Hgx9B1LdjTCTxK9CKov7hG9WbpxWxaTOA189Dy074aBE7DxwWRHI/wk0YugemO8upSpNM9BZoaSEX262fk0FNVB/UrY/0ayoxF+tid6pVSmUmqrUup5u48lrNM75CFDQYEjuhbFpowMRXm+gzYZ0aeXo5th5gUw80Jo/QCGB5IdkSAxI/qvAbsTcBxhod4hD0W52WRkqJj3UVPs5HifnIxNG33HoL8VapdD/fmgvUayF0lna6JXStUBnwSkWHeaiWdVrKmmyMmJXhnRp43jO4zbaUuhcqFxv2NP8uIRY+we0f8Y+FvAZ/NxhMWsSPTTip209g5ZFJFIeZ1Nxm3FfCiug+x8aN+b3JgEYGOiV0p9CmjTWm8Os90apdQmpdSm9vZ2u8IRUbIi0VcXO+lzj+IakdWxaaGzCZzFkFcOShkJX0b0KcHOEf2FwCql1CHgMeAypdQjp26ktX5Aa92otW6srKy0MRwRjT5/jT4e04qNRVPHpXyTHjqboHyekeQByudC18HkxiSACBO9UupJpdQnlVIR/2LQWt+jta7TWs8Cbgb+oLX+fIxxigTrHfLE3OfGVF0kiT6tdB+C0tnj3xfXQd9R8EnlNtkiTdz/BXwO2KeU+qFSaqGNMYkki7dFsWlacS4Ax/sk0U95Pp8x46a4dvyx4hngHQFXR/LiEkCEiV5r/ZrW+lagATgEvKqUWq+U+qJSKmw20Fr/UWv9qfhCFYniGvEy6ou9RbGpxj+ib5UR/dTn6jCSelHd+GPF/vu9R5ITkxgTcSlGKVUOrAa+BGwFfoKR+F+1JTKRNPE2NDPlOjIpzs3mhIzop76+o8Zt0fTxx8YSfUvi4xEniWjZo1LqKWAh8EvgOq11q/+px5VSm+wKTiRHvC2KAxlTLCXRT3m9/kR/UulGEn2qiHR9+4Na6xcDH1BK5With7XWjTbEJZLIqhE9GCdk5WRsGjA7VRYFJHpniTGXXhJ90kVauvmHII9tsDIQkTribVEcaFqxU07GpoO+o5DpgLyK8ceUMkb1UqNPupAjeqVUDVAL5CqllgFm45MiIM/m2ESSxNuiOFBNsZOOgWE8Xh/ZmdIsdcrqOwqF0yDjlP/jwmoYaEtOTGJMuNLNVRgnYOuA+wMe7wf+zqaYRJLF26I4UE2RE62hrX+Y2pLcuPcnUlT/cSPRnyq/yuhoKZIqZKLXWq8F1iqlPqO1fjJBMYkki7dFcaCasdWxQ5LopzJXF5TNnvh4QZWM6FNAuNLN57XWjwCzlFJ/ferzWuv7g7xMnOasaFFsGk/00q54SnN1QN3yiY/nV4JnEEYGwZGf+LgEEL50Y/7PFNgdiEgdVqyKNU0rMkbx0sVyCtMaXJ0nn4g1FVQZtwNtwUf8IiHClW7+23/794kJR6QCKxN9UW4WBTlZtHRLop+y3L3gGzW6Vp4q35/oB9sl0SdRpE3NfqSUKlJKZSulXldKdSilpEHZFGVloldKUVuSy9EeSfRTlqvTuM0PNqL3d6SVOn1SRTrf7UqtdR/wKaAFOAP4hm1RiaSyokVxoNrSXBnRT2WuLuM25IheEn0yRZrozZ/6a4FHtdZdNsUjUoCVI3qAutJcjna7LNufSDFmd8q8sonP5ZsjermoUDJFmuifU0p9BDQCryulKgFZ7jgFmS2K4+1FH6i2JJc+9yj9bo9l+xQpxCzdBDsZm+WA3FIZ0SdZpG2K7wZWAo1aaw8wCFxvZ2AiOaxqURyottSYeSN1+ilq0BzRByndgFG+kRp9UkWzIuZMjPn0ga/5hcXxiCSzsqGZyVwo1dI1xMKaIsv2K1KEqxOynJPPk5dFU0kXaZviXwJzgW2A1/+wRhL9lGNli2JTXanRFklG9FOUq3P8guDB5FdC67bExiROEumIvhFYpLXWdgYjks+OEX1FgYOcrAxJ9FOVqzP4iVhTfsV4HV8kRaQnY3cANXYGIlKDlS2KTeZc+haZeTM1TbYq1pRXbiyq8o4mLiZxkkhH9BXALqXU+8BY0xKt9SpbohJJY2WL4kC1pbkclbn0U9NgB5TMnPz5XP9of6h7fAGVSKhIE/290e5YKeUE3gJy/Md5Qmv93Wj3IxLLyhbFgepKc3m1tc/SfYoU4eoKvirWZJZ1XJ2S6JMkokSvtX5TKTUTmK+1fk0plQdkhnnZMHCZ1npAKZUNvK2Ueklr/W6cMQsb9QyNkJmhLGlRHKi2JJeOgRGGRrzkOsJ9dMRpY3QEhnsnn1oJ489JnT5pIu1182XgCeC//Q/VAs+Eeo02DPi/zfZ/ycncFNftMhZLWdGiOJDMpZ+ihkK0PzCZI/ohWVCfLJGejP0KcCHQB6C13gdUhXuRUipTKbUNaANe1Vq/F2ugIjF6XCOU5jss3685xVJOyE4x4RZLBT4nI/qkiTTRD2utR8xv/Iumwo7OtdZerfU5GJciPFcptfjUbZRSa5RSm5RSm9rbpR9GsnUNjlBqcX0eoL7MSPSHuyTRTylj7Q9CJPrcspO3FQkXaaJ/Uyn1dxgXCb8C+C3wXKQH0Vr3AH8Erg7y3ANa60atdWNlpZyoSbYel4eSPOtH9FWFOTizM2julEQ/pYRqUWxy5EFW7niXS5FwkSb6u4F24EPgDuBF4FuhXqCUqlRKlfjv5wKXAx/FHqpIhG6XPSN6pRT1ZXmS6KeaSEb05vOS6JMm0lk3PqXUM8AzWutI6yvTMC4snonxC+U3WuvnY4xTJIDWmm6Xx5YaPUB9WT6HuwZt2bdIEjPR54ZYGQvGCVkp3SRNuIuDK+C7wF2A8j/kBf5Na/29UK/VWm8HllkVqLCfa8TLyKiPUhtKNwAzy/N4u6kdrTVqsr4o4vQy2AHOEsgMM2bMK5dZN0kUrnTzdYzZNiu01uVa6zLgPOBCpdRf2R6dSKhul3G+3Y7SDcCs8jzcHh9t/cPhNxanB7OhWTgyok+qcIn+z4BbtNYHzQe01geAz/ufE1OI2bnSjpOxAPXlRhvbCXX6nsPQ9Dp0HbDluMJGrs7QJ2JNeeWS6JMoXI0+W2vdceqDWut2/2pXMYWMj+htKt34p1g2dw5y7uwy4+TcC38NO58e32jBtbDq3yJLHiL5XJ1QUh9+u8DGZuHKPMJy4Ub0IzE+J05DXYPGf2lZvj2/w2tLc8nMUMZcelcX/O+nYPfzcMk3YPWLcOm3YP8fjMdlhsbpIdLSTWBjM5Fw4X61LlVKBetEpQCnDfGIJLK7dJOdmcH0EifNHYPw9Legcx/c+luYe6mxwawLYca58Mhn4Hd3wc2/mvxiFiL5tDZOxkZaowdpbJYkIUf0WutMrXVRkK9CrbWUbqYYs3Rj5YXBTzWzLJ95rb+Dfa/Ald8fT/KmOR+Dy++FPS/A7mdti0NYYLgffJ4IE71/G5l5kxSRLpgSaaB7cIRCZxZZmfZ9LOaWKG7pXwt158K5Xw6+0Xl3QtUiePU7MCozdFKWy3/6LqKTsdIGIZkk0Ysx3S4PZTYtljJ9cvAZKulm4GP3Tl6WycyCK/8Bug/B1l/aGo+IgyuCzpUmaWyWVJLoxZhu14ht9XkARodZ2vob3vQuocl5Vuht514Gtcth/b+Dzxt6W5EcY+0PIhjRjzU2k9JNMkiiF2N6XB7bFksBsPNpctztPOS9hv1tA6G3VQou/Bp0H4TdEffPE4k01qI4TPsDCGhsJiP6ZJBEL8YYLYptHNFvfAhdfgbrWcr+9jCJHmDhp4w52hsftC8mEbtIOlcGksZmSSOJXozpcdmY6LsOQMv7qGWfY2Z5PgfaI2hulpEJDbfBoXXQ0WRPXCJ2rg7IdICjILLt88pk1k2SSKIXAAyPehkc8dpXutn+W+P27M8yp7IgshE9wLLPQ0YWbPlfe+ISsTMXS0W61kH63SSNJHoBQOeAMYe+ojDH+p1rDR/+BmZeBMV1zK0s4FDnIKNeX/jXFtbAgmtg269lqmWqGeyM7ESsSfrdJI0kegGMJ/pyO6ZXHt8OnU2w5LMAzK3Mx+PVHOmO8ELhy1cbCUJOyqYWV2dkJ2JNkuiTRhK9AKBjwBgt2zKi3/MSoIyTq8DcKqOmG3bmjWnOZVAyEzb9j/WxidhF2rnSNNbYzGNfTCIoSfQCCEj0+TYl+roVY0lhboWR6A90RJjoMzKg8YvQ/Da0ydUoU4Yrwj43prE2CNLYLNEk0QsAOsZq9BaXbvpaoXUbLBi/LnxxXjYVBQ72t0VxWcFlXzBmeGx62Nr4RGy8HmN0Hkuil/JNwkmiFwB0DgyTm51JnsPiXuH7XjZuz7jmpIejmnkDxl8Di66HDx6FEbnubNJF0/7AZG47OOESF8JmkugFAJ2DI5QX2HAidt+rUFwPVWee9PDcygKa2gfQWke+r8bbYbgPPnzC4iBF1MbaH0SR6M16vozoE04SvQCMGn1FgcX1eZ8PmtfDnEsmzLU+o7qAHpeH9miuH1t/PlSfDRv+w9i3SJ5oOleaxko3MqJPNNsSvVJqhlLqDaXUbqXUTqXU1+w6lohfx8AIFVaP6Dv2GCsh6y+Y8NSCmkIA9pzoj3x/SsFFXzf2+9HzVkUpYhHLiF4amyWNnSP6UeBvtNZnAucDX1FKLbLxeCIOnXaM6JvfMW5nBkn01f5EfzyKRA9w1g1QNgfW/YuxEEskRzSdK01ZDsgpltJNEtiW6LXWrVrrLf77/cBuoNau44nY+Xzanhp983oonA6lsyY8VV6QQ0VBTvSJPiMTLv4baP1ArkCVTINmoo9iwZS5vZyMTbiE1OiVUrOAZcB7iTieiE7vkAevT1Nu5Rx6raF5gzGan6QXyoKaguhKN6alt/ivQPVdGJVr1CeFq8MYnWdG2Rspv0JG9Elge6JXShUATwJf11pPuNC4UmqNUmqTUmpTe3u73eGIIDoHbVgV230I+o8FLduYFlQXsfdEPz5flCWYjEy48j6jV/3Gn8cXp4jNYEd0J2JN0gYhKWxN9EqpbIwk/yut9VPBttFaP6C1btRaN1ZWytXhk6Gt31wVa2Hppnm9cTvzwkk3WVhTiNvj43CXK/r9z7vcuArVmz+Sk3vJMNgO+TH8vEqiTwo7Z90o4CFgt9b6fruOI+J3os8NQHWx07qdNq83ZllULph0kzP8M28+irZOb7riPmN15js/ie31InbR9rkxmYleTqQnlJ0j+guBLwCXKaW2+b+utfF4IkYn+owRfXWRlYn+nZD1eTDm0gPsjaVOD1CzGJb8Kbz3M+g7Fts+RGziKd2MumV1c4LZOevmba210lov0Vqf4/960a7jidgd73VTkJNFQY5F7Q/6jhn18xD1eYA8Rxb1ZXnRz7wJ9PF7jIuHv/lPse9DRMfn87cojiHRy+rYpJCVsYK2fjfVRRaeiB2rz4dO9GAsnIpp5o2pbDYsvw22PmI0UBP2c/eA9sZeowdJ9AkmiV5wom/Y4rLNenAUGu0KwlhQXcjBjkHcHm/sx1v5FWNUL50tE2PQPzsuptKNjOiTQRK94Hiv29pEf3gD1J8HmeFLQWdOK8Lr0+w7EUUny1OVzYEzrjYSvVxu0H6DMfS5MZkLrCTRJ5Qk+jSntfaXbixK9K4uaNsVUdkGYHFtEQA7jvXGd9wVXzIW8ex7Jb79iPDMEX0sNXppVZwUkujTXNfgCB6vtq5Gf3iDcRukkVkwM0rzKMzJYme8iX7Ox42a8fbH49uPCG+sc2UMNXpnMWRkyYg+wSTRpzlzamWNVSP65vWQmQO1DRFtnpGhOHN6ETuPTVg0HZ3MLFh8I+x9GYZ64tuXCM0cjUfb5waM6bayaCrhJNGnOXOxVJVlif4d4/qwWZH/hbB4ejG7W/vwRtsK4VRnfxa8I/6LkQvbDHZAbmn0fW5MedLvJtEk0ae5sVWxVpRuhvuNrpIR1udNZ00vwu3xcSCaSwsGM30ZFNTA3t/Htx8R2mB7bPV5U37FeJ1fJIQk+jRnlm6qCi0Y0R95D7Qv+kTvPyEbd/kmIwPOuBKaXpeulnZydcZWnzcVVMPACeviEWFJok9zx/vclOc7cGRZ8FFoXm+caJtxblQvm1dZQE5WBjuOxnlCFoyLkI/0w+H18e9LBDfYDvlRXFnqVAVVMNAm/W4SSBJ9mmvrc1tYn98A05aCIz+ql2VlZrCwpjD+ET0Ys2+ynLBHyje2GeyIf0TvccFInKU6ETFJ9GnuaM8Q063oWulxw9FNUZdtTIumF7PzWC863lGeIw9mXQz7Xo5vPyI4nzf2PjemgmrjdqDNmphEWJLo09zRniFqS3Mt2NFmY8ZLiP7zoSyuLaLPPcqRrqH4Y5l/BXQdgM798e9LnGywA9BG+SVW5mv7j1sSkghPEn0a63N76HePUltiQaJvXg8oqD8/ppcvrSsBYOuR7vhjmXe5cbv/D/HvS5ys3984rnBa7PsYG9HLCdlEkUSfxo52G6NnS0b0ze9A9VnG/OoYLKwpJM+RyZZmCxJ9+VwonQ1Nr8W/L3EyMzkX1sS+D/b5PbUAABadSURBVCndJJwk+jQ2lujjHdF7PXDk/Zjr82CckD1nRgmbD1uQ6MEY1R98yzh3IKwzNqKPI9Hnlhqzs2REnzCS6NNYS7dxrda60rz4dtS6HTyDUL8yrt00zixld2s/g8Oj8cUDRp3e4xrvvSOsYdbVzVF5LDIyIL9KRvQJJIk+jR3tGSInK4OKgjgvCt78jnEbx4geoGFmKV6f5oMjFvSqmXURZDqkfGO1/uPGjJtY2x+YCqpkRJ9AkujT2NGeIWpLclEhrusakeb1UDY3vj/ngWX1Rn1/sxV1eke+8YtHEr21+o/HdyLWJKtjE0oSfRo72m3B1Eqf11iFGudoHqA4N5sF1YW8f6gr7n0BRp2+/SPoOWLN/oRRoy+Mo2xjkhF9QtmW6JVSDyul2pRSO+w6hoiPOaKPy4kd4O6F2ZdYEtOF8yp4/2BXfJcWNM27wrjd/3r8+xKGgRNx/+UGQHGdUaOXnkQJYeeI/n+Bq23cv4iD2+OlY2Ak/kR/6G3jNsaFUqe65IwKhkd9vHfQglF95QIoqpPyjVV8Xn+it6B0U1QLaOg/Fv++RFi2JXqt9VuARX+DC6sd7TGmVtaVxZvo3zHmrBfXWhAVnDe7HEdWBm/ttaCNrVIw7xNw4E1jCqiIz2C70Z00nhk3JvPz0ns0/n2JsKRGn6ZaxubQxzG10uczZtzMusiiqCDXkcl5s8v44x6Lpt7NuxyG+4x5/iI+VsyhNxXPMG57W+Lflwgr6YleKbVGKbVJKbWpvV0uRpAozZ2DAMwqjyPRn9gB7h6jiZiFPrGwiv3tgzS19ce/szkfMxbnSPkmfmZSNpN0PIr8I/o+SfSJkPREr7V+QGvdqLVurKyMo/WpiMqhDhe52ZlUFsZxZSmzPj/Lmvq86Zqzp6EUvLDdgqZXzmKYcZ4keiv0HDZuS+pDbra5uYs3w5XeHHnGClkp3SRE0hO9SI7mzkFmlufFN4f+0Nv++nyddYEB1UVOVswq44UPLTpRN+8TcHw79LVas7901XMEsvND9jN670AnN/5sA7c9/D6/2xYmiRfXQZ8k+kSwc3rlo8AGYIFSqkUpdbtdxxLRO9Q5yKzy6C4QchLvqJHoZ1tbtjF9ask09p4YYO8JC8o3Z64ybrc/Hv++0lnvESiZYZzknsS/v9FEWZ6DORX53P/qXnyhLvheVCc1+gSxc9bNLVrraVrrbK11ndb6IbuOJaLj9WmOdA0xsyKO+vzRTTDcC3M/YV1gAa5eXEOGgme3WTCqr5hvTP/cstY4gSxi03M4ZNnmRJ+bdfs6+MLKmXzl0nk0d7rYHurykMW1kugTREo3aai1d4gRry++EX3Ta6AyjUv32aCq0MlF8yt5aktL6FFhpJavNi5GcvCP8e8rXfUeCXki9o2PjJlSVy+u4fIzq8nKULyyM8R5luI642T+sFxS0G6S6NNQc6fRtXJmPDNuml6DuhWQW2JRVBN9dnkdx3rdrN/fGf/OzlwFhdPhD9+Xi1LHYngAhrqN0s0k3m7qoKbIyYLqQorzsllSV8y7B0L835XMNG67D1kbq5hAEn0aOthhTq2McUQ/2AHHthknOW10xaJqipxZPLHZgl412U649O+MktPGB+PfX7rpOmDcls6adJMtzd0sn1U6doJ/xewyPjzaO3k7i/K5/n3LJR/tJok+DR1oH8SZnUFNUYwXBd//BqBtT/TO7ExWnTOdl3Ycp89twcrWc26F+VfB7++B9/4bRgbj32e66GwybsvnB336eK+bY71ultePz8g5d1YZHq9m22Rtp8vm+Pctid5ukujT0L62fuZXFZKREePUyn0vQ145TDvH2sCCuHH5DIZHfbyw3YKpkRkZ8OkHjAZsL/0t/ON0uK8Kvj8dflgPv74ZjksPvqDMZGwm51Ns8V8ZrGHmeKJvnFmGUrBpsm6kOYXGBUhkRG+7rGQHIBJvz/F+LppfEduLR4dh78uwaBVkZFobWBBL64qZX1XAbzcd4ZZzQy/UiUhuCdz6BBxaZ7RFGOk3mnWNDMLuZ+GBj8Nn/wfOvC7+Y00lnU3GdEhH8PM6W5q7ycnKYNG0orHHivOymVdZwJbDIS4kUz4XOg9YHa04hST6NNPr8tDWP8wZ1YWx7eDAm0bvGHNuus2UUty4vI4fvPQRTW39zKuKMe5AGRlGa4Q5Hzv58cu+DY/eBL9dDX/+MtQ1xn+sqaKzabymHsSWw92cXVuMI+vkIsGy+hJe3XUCrXXwxXllc6HpVaujFaeQ0k2a2evvH3NGdUFsO9j9LOQU2TatMpjPLK/DkZnBLzY023ug/HK49bfG7Jwnvmj02RfGLKXOfVA+L+jTHq+PHcf6WFY/cQbWsvpSul2esZleE1TMN1ofu6TRrZ0k0acZc6Xp/FhGxt5R2PMinHEVZMXRIydKFQU5XLd0Ok9sbrHmpGwouaVw40PGcv83fmDvsU4XfceMX3qVC4M+ved4PyOjPpbOmJjoG/wnZ80a/gQ1i43btl2WhCqCk0SfZvadGCDPkRnbBUf2vw6uTjjrBusDC2P1BbNwjXj57aYErKScca6xwOr9B6Btt/3HS3XHPzRupy0J+rQ5q2Zp3cREP6+qgIKcLLZOVqev9id6OQluK0n0aWZXax/zq2OccbP1l5BfCfOvtD6wMM6uK2bFrFIeXHfAmssMhnPZt41ZIS/9X1lgdfxDQEH1WUGf/uBID2X5DuqCXH84M0OxdEbx5CP6gmpjBtcJSfR2kkSfRrw+zY6jvSytK47+xYMdsOf3sOQmyMy2PrgI/NUVZ9Da6+aRd22u1YNRr//4PXDwTdj7e/uPl8qObzemVeYEL/d90NLD0rriSTuhNtSX8tHxflwjoxOfVMoY1Uuit5Uk+jSyv30A14g36J/YYW1/HHweY9FRklwwt4KL5lXw72800d4/bP8BV9xuLBB65VvpfRHr1m1Qc3bQpwaGR9nXNhC0Pm9aVl+C16f5sGWSk9vTlxmlG8+QFdGKICTRp5GxWmqIH8qgvKPGStIZ50H1Ihsii9y9qxbhGvHy7Wd2oO0uqWRmw5X/YEwt3JSmzVd7jxpdK+vPD/r09pYetA79mVo2wzwhO0mdvn6lMYg4uiXucEVwkujTyLYjPRTmZDGnIsoeNx89Bz3NcMFf2hNYFOZVFfLXV5zB73cet3+6JRgzjOZ8HP74w/ScAtj8jnE784KgT2882I1SsCxEoi/NdzC7Ip+tk9XpZ5xr3B7eEE+kIgRJ9Glkw/5OVswui+5ErM8H6+43arQLrrUvuCisuXgOVyyq5nvP7+KtcJesi5dScNU/GovE3vwne4+ViprfgZzi8dkxp9hwoINF04ooyXOE3M2yGSVsPdIT/K+wvDKoPHP8l4qwnCT6NNHS7eJgxyAXzouy9cGOJ42TcR+/JyEtDyKRkaH415vOYX5VAXc+spnNzZOMFK1SfRY03Abv/xxaNtt7rFSiNTS9DrMuCvp/7/Z42XK4h5VzysPuatnMUtr7h2npnqQOP+8TxhXL3H3xRi2CkESfJt5p6gDg4mh63AwPwOvfg+qzYfGNNkUWm4KcLH7x5+dSXeRk9cPvT36izyqX3wtF0+HJ22EoRO+WqeToFuNiI5P0/dl4qIuRUR8r54ZP9OfPLgOY/KLhCz8J3hG5iLtNJNGnid/vOE5tSS7zq6JoffDavcYP+rU/MvrDpJiqIie/+tJ5FOVm87kH32X9/g77DpZbAp/+uXHpu0dvgZFJlvRPJbuehoxsWHB10Kdf/PA4eY7MiP5KnFdVwOyKfF6e7IpTM86DvArY+VQ8EYtJpN5Pr7BcW7+bt/Z1cP050yed6zzBrmdh48/hvDsnPRGXCqaX5PKbO1dSU+Rk9cMbefjtg9ZcejCYmSvh0/9tnDR8+Epo32vPcVLByCBs+aWR5HNLJzw96vXx8s7jXLawCmd2+JKeUoorF1WzYX8nPa4gU1UzMmHZrfDRi3IdWRtIok8DT24+iten+XRDbWQvOPwuPH0H1DYaJYsUV1uSyxN3XsBF8yv43vO7uPFn6/njnjZ7Ev7iz8DnHjemHP7nefD452HzWuM96242LrfnS8DKXbtt/l/jeq4rg8+0+sNHbXQNjvCpJdMj3uX159Qy6tM8tnGSK4at+BKg4Z2fRh+vCMnWNsVKqauBnwCZwINa6x/aeTwxUZ/bw4PrDnDx/IrIWvzuetZI8oXT4OZfG5fgOw0U52Xz0G2N/HZzC/e/spfV/7ORmiInV51VzaULqzhvdjm5DotOJp9xFdy1Gdb/BLb/BnY/N3EbRyE4i42Sj7MYnCXBvy+ZYZzsDTJqThqzodvcy6D+vAlP+3ya//zjfmpLcrn8zKqId7toehEXzC1n7fpDrL5g1sS/BErqYfkXjb8kl94MtQ3x/kuEn7Jr0YlSKhPYC1wBtAAbgVu01pO2qWtsbNSbNm2yJZ505PNpvv74Np7ffoxnvnIhS0KtiO3YB2/8o1Ejnd5gjFoLIv8hTiUjoz5e2tHKix+28ubedtweH47MDFbMLuWieZVcPL+CRdOKYr/CViCfD3oOQUeT0W53uM+YOTLcZ3R8HOoxbt094/dH+ifup6jOWH0a+FU6y5jemUh9rfDLG4xzM3e+DWWzJ2yydv0hvvvsTn504xL+tHHyi4UH8+6BTm5+4F3+bOVM/n7VWRNLia4u+NlFoH3wZ7+DygXx/GvSglJqs9Y65MUT7Ez0K4F7tdZX+b+/B0BrPWnvV0n01nB7vOw81stPXm/irb3tfOOqBXzlUn8vca2NJDTYYVwe7thW2P8HOPIuZOfByrvgkm9AVuh50acLt8fL+we7WLevnXX7OvjouJFky/MdnD+nnKqiHIpzsynIySLXkUmeI5M8RxaFziyKc7MpyXNQnJtNviMz8vMb4XhHjf+DoW7oOggnPjRaAJzYAR17jSQHxl8FxXVQWG1ccs9ZBI58cBQY/1fmfUe+ceWn7DzIzg24zYWsXKOl9GSxa2388unYZ1wA5P0HjKuIfe5x45KLY5vpsT5D//Xmfi5dUMWDf9YY0y/L7z23i4ffOchnl9dxx8fmMrcy/+T39sROWLsKPC5jhH/Wnxgtkp1Fk+80jSU70d8IXK21/pL/+y8A52mt75rsNbEm+k/92zrcHt9JizHMe/85+Dc4tZvAj6Mae1YHPDb+nD5lO8XE92jsMR1+f+P0Sc9FfIxTthuPL/h+zLdBAc7sDByZCmVuPTpkTGMLjLT6LFj0J7D8ttN2FB+ptj43bzd18Pa+DjY2d9E96GFgOEizrVNkZSgyMwL+R4L92KiJd9VJj6mTHjvp/8//oJMR5nGYheoQZ3CYKrqopJtyesljiDzcOImuJ78PhZscvEFOyTnwkBOwv/WqgZ9mraY5o24sRg30uDwM+buGfnpZLd+/4eyYS2E+n+ZHL+/h5+sO4PVpnNkZlOQ6cGZnoJTxLlXoTr48vJbLRteRifGLz0UubpWDh2xGVVbQf8+44L+A9KSPJ8cfsj/Gozl/Smmeg9/cuTKmfUSS6O2s0Qd7Rye8n0qpNcAagPr62K4JOq+yAI83ILsFBNDbOhvX2Ad5/EmN+SEeS4FopSZsBwrUqSk8yD/N/1rzg6QC7gcbTWkV7EMaGPz4/tQpj520rQr8Nxn3HZkZlOQ7mF6SS05W5snbZzmMVsP5lVA8w+gxPklXwqmoqsjJpxvq+HRD3dhjo14fLo+XoREvrhEvg8Oj9LtH6R3y0Ds0Qu+Qhx6XB682f5menLDh5E7GOmAQMP6YuZ0Osv2p+5iLRrMP2BskAymfh2yfm2zvEA6fC4d3CIdviCzfsPG4zx1w33+r3WSYfykE8JDNQFYp3dk1HMxfymBWKfVo6gPi0UBxbjZ1pblcPL+SedFM0Q0iI0Nx9zUL+cLKmazb205T2wB9bo8xWBt7j4p4hnt5zdPDXPd2KjxHKR7txOEbJkt7yNKeoIMjCD5oGv+XRLO9/bLzpjG/uIAip70dYaV0I4QQp7FIRvR2Tq/cCMxXSs1WSjmAm4FnbTyeEEKIIGwr3WitR5VSdwEvY0yvfFhrvdOu4wkhhAjO1nn0WusXgRftPIYQQojQZGWsEEJMcZLohRBiipNEL4QQU5wkeiGEmOIk0QshxBRn24KpWCil2oEEXPE5KhWAjVe0sITEaJ3TIU6J0RpTJcaZWuvKUBukVKJPRUqpTeFWnSWbxGid0yFOidEa6RSjlG6EEGKKk0QvhBBTnCT68B5IdgARkBitczrEKTFaI21ilBq9EEJMcTKiF0KIKU4SPaCUKlNKvaqU2ue/nXClZqXUDKXUG0qp3UqpnUqprwU8d69S6qhSapv/61oLY7taKbVHKdWklLo7yPNKKfVT//PblVINkb42gTHe6o9tu1JqvVJqacBzh5RSH/rfN9suRhBBjB9XSvUG/B9+J9LXJjDGbwTEt0Mp5VVKlfmfS9T7+LBSqk0ptWOS51Ph8xguxlT4PIaL0drPo9Y67b+AHwF3++/fDfxTkG2mAQ3++4UYFz5f5P/+XuD/2BBXJrAfmAM4gA/MYwZscy3wEsYlpM4H3ov0tQmM8QKg1H//GjNG//eHgAqb/38jifHjwPOxvDZRMZ6y/XXAHxL5PvqPcwnQAOyY5Pmkfh4jjDGpn8cIY7T08ygjesP1wFr//bXAn5y6gda6VWu9xX+/H9gN1Noc17lAk9b6gNZ6BHjMH2ug64FfaMO7QIlSalqEr01IjFrr9Vrrbv+37wJ1JFY870XKvI+nuAV41IY4QtJavwV0hdgk2Z/HsDGmwOcxkvdxMjG9j5LoDdVa61YwEjoQ8grZSqlZwDLgvYCH7/L/KfhwsNJPjGqBIwHftzDxl8tk20Ty2kTFGOh2jBGfSQOvKKU2K+P6wXaINMaVSqkPlFIvKaXOivK1iYoRpVQecDXwZMDDiXgfI5Hsz2O0kvF5jJRln0dbLzySSpRSrwE1QZ76ZpT7KcD4Afu61rrP//B/AfdhfEjuA/4f8OexRzt+uCCPnTpNarJtIro4uwUiPo5S6lKMH6yLAh6+UGt9TClVBbyqlPrIP9pJdIxbMJaSD/jPsTwDzI/wtVaI5jjXAe9orQNHhIl4HyOR7M9jxJL4eYyEpZ/HtBnRa60v11ovDvL1O+CE/89L/LdtwfahlMrGSPK/0lo/FbDvE1prr9baB/wc488rK7QAMwK+rwOORbhNJK9NVIwopZYADwLXa607zce11sf8t23A01j33kUVo9a6T2s94L//IpCtlKqI5LWJijHAzZxStknQ+xiJZH8eI5Lkz2NYln8e7T7pcDp8Af/MySdjfxRkGwX8AvhxkOemBdz/K+Axi+LKAg4Asxk/8XLWKdt8kpNPfr0f6WsTGGM90ARccMrj+UBhwP31wNVJirGG8XUl5wKH/e9pyryP/u2KMWq7+Yl+HwOON4vJTyIm9fMYYYxJ/TxGGKOln0db/gGn2xdQDrwO7PPflvkfnw686L9/EcafSNuBbf6va/3P/RL40P/cswQkfgtiuxZjhs9+4Jv+x+4E7vTfV8B/+J//EGgM9Vqb3r9wMT4IdAe8b5v8j8/xf1A/AHYmOca7/DF8gHGC7oJQr01GjP7vV3PKQCLB7+OjQCvgwRhd3p6Cn8dwMabC5zFcjJZ+HmVlrBBCTHFpU6MXQoh0JYleCCGmOEn0QggxxUmiF0KIKU4SvRBCTHGS6IUQYoqTRC+EEFOcJHohhJji/j8b7fWGmK0GwgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_df.groupby(\"class\")[\"domain_consonant_variety_ratio\"].plot.kde(legend=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "7QqEFPESP7nS"
   },
   "source": [
    "## Character Pair Windows\n",
    "\n",
    "A typical step in natural language processing is to tokenize the documents, and then convert the tokenized representation into a numerical representation of the document e.g. Bag of Words, n_grams, Word Vectors.\n",
    "\n",
    "Let's take a slightly naive approach and create a sliding 2 character window over our domains, from this we can find the most frequently occuring two character pairs across the classes. \n",
    "\n",
    "The rationale is that in languages there are certain characters that will frequently appear alongside another. We will build this window in a left to right fashion which does present a limitation for right to left languages.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "JSv_3a6cP7BE"
   },
   "outputs": [],
   "source": [
    "from itertools import islice\n",
    "\n",
    "def text_window(seq, n=2):\n",
    "    \"Returns a sliding window (of width n) over data from the iterable\"\n",
    "    \"   s -> (s0,s1,...s[n-1]), (s1,s2,...,sn), ...                   \"\n",
    "    it = iter(seq)\n",
    "    result = tuple(islice(it, n))\n",
    "    if len(result) == n:\n",
    "        yield result    \n",
    "    for elem in it:\n",
    "        result = result[1:] + (elem,)\n",
    "        yield result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "4dlhO9S41y_A"
   },
   "source": [
    "Now let's bring the class back in"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "WtSy6IBARZic"
   },
   "outputs": [],
   "source": [
    "domain_windowed = data_df[\"domain\"].apply(lambda x: [\"\".join(y) for y in text_window(x, 2)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 119
    },
    "colab_type": "code",
    "id": "8x7VWOf618vP",
    "outputId": "cdf8b844-f338-486c-8af4-19ca3a49d1f0"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0         [00, 00, 0d, di, ir, re, ec, ct, to, or, ry]\n",
       "1                 [00, 00, 0w, we, eb, bh, ho, os, st]\n",
       "2                             [00, 01, 1f, fa, an, ns]\n",
       "3    [01, 1-, -t, te, el, le, ec, ch, ha, ar, rg, g...\n",
       "4                 [01, 10, 0s, sh, ha, an, ng, gp, pu]\n",
       "Name: domain, dtype: object"
      ]
     },
     "execution_count": 122,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domain_windowed.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "_9zNQtu417fC"
   },
   "source": [
    "Let's encode the above into a sparsely populated dataframe in a similar fashion to one hot encoding."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 270
    },
    "colab_type": "code",
    "id": "hK-qqVmMReaf",
    "outputId": "0321b677-7185-4d55-ead4-cfdd88fd5844"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>\\nd</th>\n",
       "      <th>--</th>\n",
       "      <th>-0</th>\n",
       "      <th>-1</th>\n",
       "      <th>-2</th>\n",
       "      <th>-3</th>\n",
       "      <th>-4</th>\n",
       "      <th>-5</th>\n",
       "      <th>-6</th>\n",
       "      <th>-7</th>\n",
       "      <th>...</th>\n",
       "      <th>zq</th>\n",
       "      <th>zr</th>\n",
       "      <th>zs</th>\n",
       "      <th>zt</th>\n",
       "      <th>zu</th>\n",
       "      <th>zv</th>\n",
       "      <th>zw</th>\n",
       "      <th>zx</th>\n",
       "      <th>zy</th>\n",
       "      <th>zz</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 1368 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   \\nd  --  -0  -1  -2  -3  -4  -5  -6  -7  ...  zq  zr  zs  zt  zu  zv  zw  \\\n",
       "0    0   0   0   0   0   0   0   0   0   0  ...   0   0   0   0   0   0   0   \n",
       "1    0   0   0   0   0   0   0   0   0   0  ...   0   0   0   0   0   0   0   \n",
       "2    0   0   0   0   0   0   0   0   0   0  ...   0   0   0   0   0   0   0   \n",
       "3    0   0   0   0   0   0   0   0   0   0  ...   0   0   0   0   0   0   0   \n",
       "4    0   0   0   0   0   0   0   0   0   0  ...   0   0   0   0   0   0   0   \n",
       "\n",
       "   zx  zy  zz  \n",
       "0   0   0   0  \n",
       "1   0   0   0  \n",
       "2   0   0   0  \n",
       "3   0   0   0  \n",
       "4   0   0   0  \n",
       "\n",
       "[5 rows x 1368 columns]"
      ]
     },
     "execution_count": 123,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.preprocessing import MultiLabelBinarizer\n",
    "\n",
    "mlb = MultiLabelBinarizer()\n",
    "\n",
    "domain_windows_df = pd.DataFrame(mlb.fit_transform(domain_windowed), columns = mlb.classes_, index=data_df.index)\n",
    "domain_windows_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "epBRQ6n_2r-9"
   },
   "source": [
    "Let's being class back in."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "8X2AcgsNRvN0"
   },
   "outputs": [],
   "source": [
    "domain_windows_columns = list(domain_windows_df.columns)\n",
    "domain_windows_df[\"class\"] = data_df[\"class\"]\n",
    "domain_windows_df = domain_windows_df[[\"class\"] + domain_windows_columns]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 270
    },
    "colab_type": "code",
    "id": "_mNmS6AFozQz",
    "outputId": "b1b504be-913b-4a64-8b29-db0de0930881"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>class</th>\n",
       "      <th>\\nd</th>\n",
       "      <th>--</th>\n",
       "      <th>-0</th>\n",
       "      <th>-1</th>\n",
       "      <th>-2</th>\n",
       "      <th>-3</th>\n",
       "      <th>-4</th>\n",
       "      <th>-5</th>\n",
       "      <th>-6</th>\n",
       "      <th>...</th>\n",
       "      <th>zq</th>\n",
       "      <th>zr</th>\n",
       "      <th>zs</th>\n",
       "      <th>zt</th>\n",
       "      <th>zu</th>\n",
       "      <th>zv</th>\n",
       "      <th>zw</th>\n",
       "      <th>zx</th>\n",
       "      <th>zy</th>\n",
       "      <th>zz</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>legit</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>legit</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>legit</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>legit</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>legit</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 1369 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   class  \\nd  --  -0  -1  -2  -3  -4  -5  -6  ...  zq  zr  zs  zt  zu  zv  \\\n",
       "0  legit    0   0   0   0   0   0   0   0   0  ...   0   0   0   0   0   0   \n",
       "1  legit    0   0   0   0   0   0   0   0   0  ...   0   0   0   0   0   0   \n",
       "2  legit    0   0   0   0   0   0   0   0   0  ...   0   0   0   0   0   0   \n",
       "3  legit    0   0   0   0   0   0   0   0   0  ...   0   0   0   0   0   0   \n",
       "4  legit    0   0   0   0   0   0   0   0   0  ...   0   0   0   0   0   0   \n",
       "\n",
       "   zw  zx  zy  zz  \n",
       "0   0   0   0   0  \n",
       "1   0   0   0   0  \n",
       "2   0   0   0   0  \n",
       "3   0   0   0   0  \n",
       "4   0   0   0   0  \n",
       "\n",
       "[5 rows x 1369 columns]"
      ]
     },
     "execution_count": 125,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domain_windows_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "BRKgxghU2yYV"
   },
   "source": [
    "Now we'll melt aggregate and melt the dataframe above so we have summed totals."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 235
    },
    "colab_type": "code",
    "id": "hMzQHQuPRkxv",
    "outputId": "755b6bf8-1e04-4840-f704-f2ded1a8b720"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>class</th>\n",
       "      <th>variable</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"5\" valign=\"top\">legit</th>\n",
       "      <th>in</th>\n",
       "      <td>12805</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>er</th>\n",
       "      <td>12336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>re</th>\n",
       "      <td>10424</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>an</th>\n",
       "      <td>9890</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>es</th>\n",
       "      <td>9260</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                value\n",
       "class variable       \n",
       "legit in        12805\n",
       "      er        12336\n",
       "      re        10424\n",
       "      an         9890\n",
       "      es         9260"
      ]
     },
     "execution_count": 126,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "domain_windows_count_by_class_df = domain_windows_df.groupby([\"class\"]).sum()\n",
    "domain_windows_count_by_class_df = domain_windows_count_by_class_df.reset_index().melt(id_vars=\"class\").groupby([\"class\", \"variable\"]).sum().sort_values(\"value\", ascending = False)\n",
    "domain_windows_count_by_class_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "DAhlPjp63Gbr"
   },
   "source": [
    "For the purposes of visualising we'll need a count of total records by class type."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Az59InGVzDVX"
   },
   "outputs": [],
   "source": [
    "total_class_counts = data_df[\"class\"].value_counts()\n",
    "total_class_counts.name = \"total\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "jgGcinpr3BOA"
   },
   "source": [
    "As we increase the number of pairs we consider the greater the likelihood of getting matches in both the `legit` and `dga` class.\n",
    "\n",
    "Let's gradually increase the number of pairs we're considering and plot the matches compared the total count per class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "JpOnf31t3vcZ"
   },
   "outputs": [],
   "source": [
    "def display_matching_pairs(windows_count_by_class_df, domain_class, max_pairs, total_class_counts, data_df, domain_windows_df):\n",
    "    matches = list()\n",
    "\n",
    "    grid_col_size = 4\n",
    "\n",
    "    fig, ax = plt.subplots(\n",
    "        figsize=(5 * grid_col_size, 4 * (max_pairs // grid_col_size)), \n",
    "        ncols = grid_col_size, \n",
    "        nrows = (max_pairs // grid_col_size)\n",
    "    )\n",
    "\n",
    "    character_pairs = windows_count_by_class_df.loc[domain_class, :].sort_values(\n",
    "        \"value\", ascending = False\n",
    "    ).head(max_pairs).reset_index()[\"variable\"].values\n",
    "    \n",
    "    print(f\"top {max_pairs} character pairs for {domain_class}:\", character_pairs)\n",
    "\n",
    "    for i in range(0, max_pairs):\n",
    "        class_matching_character_pairs = data_df[(domain_windows_df[character_pairs[:i+1]] > 0).apply(any, axis = 1)][\"class\"].value_counts()\n",
    "        class_matching_character_pairs.name = f\"{i+1}\"\n",
    "        matches.append(class_matching_character_pairs)\n",
    "\n",
    "        pd.DataFrame( [ total_class_counts, class_matching_character_pairs] ).T.plot.barh(legend=\"True\", ax = ax[i // grid_col_size, i % grid_col_size])\n",
    "\n",
    "    return pd.DataFrame(matches)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "u_Wo2t3T4wiJ",
    "outputId": "b51eeb8f-6814-4a37-f82a-c4e6402416ed"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "top 20 character pairs for legit: ['in' 'er' 're' 'an' 'es' 'ar' 'on' 'or' 'st' 'ne' 'te' 'li' 'al' 'ra'\n",
      " 'en' 'ma' 'me' 'to' 'ti' 'le']\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 20 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "legit_pair_matches_df = display_matching_pairs(domain_windows_count_by_class_df, \"legit\", 20, total_class_counts, data_df, domain_windows_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "iHG0ozXj-aa4"
   },
   "source": [
    " Lets we visualise the growth as we increase the number of pairs as a line plot."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "r_yO1zPlH1hU"
   },
   "outputs": [],
   "source": [
    "legit_pair_matches_df[\"diff\"] = legit_pair_matches_df[\"legit\"] - legit_pair_matches_df[\"dga\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 282
    },
    "colab_type": "code",
    "id": "s6w1a3qe5j8O",
    "outputId": "e2435a3e-7ee1-485b-db77-28da53083db3"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2091f712278>"
      ]
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "legit_pair_matches_df.plot(legend=True, ylim=(0,len(data_df)), xticks=range(0,21))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "28VewWwiBEtz"
   },
   "source": [
    "The change in `dga` matches seems to be almost linear, let's investigate the rate of change as a gradient to identify where the plateau effect kicks in."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 265
    },
    "colab_type": "code",
    "id": "OQbaMW2DB6al",
    "outputId": "e844573f-55eb-4009-a6a1-8352da2cbb7c"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = (legit_pair_matches_df.apply(np.gradient)).apply(pd.Series).plot(\n",
    "    xticks=np.arange(20)\n",
    ")\n",
    "ax = ax.set_xticklabels(list(range(1,21)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "iyiBy5ZuBj5Q"
   },
   "source": [
    "It would seem the plateau kicks in good and proper from 9 pairs onwards. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "b5aQHHIyJlRm"
   },
   "source": [
    "Let's review the above but for the top pairs for `dga` ."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "D-Nn45Yk6PLe",
    "outputId": "4f38a614-4041-4d3d-8abb-4dd6fd214d11"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "top 20 character pairs for dga: ['yh' 'kb' 'yp' 'cu' 'yt' 'yl' 'yx' 'ws' 'qk' 'yd' 'fu' 'oj' 'rc' 'uk'\n",
      " 'nb' 'hu' 'lj' 'mr' 'tk' 'lr']\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 20 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "dga_pair_matches_df = display_matching_pairs(domain_windows_count_by_class_df, \"dga\", 20, total_class_counts, data_df, domain_windows_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "4dumBLx5ICrl"
   },
   "outputs": [],
   "source": [
    "dga_pair_matches_df[\"diff\"] = dga_pair_matches_df[\"dga\"] - dga_pair_matches_df[\"legit\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 282
    },
    "colab_type": "code",
    "id": "hPWFs9nFIwgt",
    "outputId": "ede49a43-c65a-467d-d6e1-4cf68877aea2"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x209205a1278>"
      ]
     },
     "execution_count": 135,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYoAAAD4CAYAAADy46FuAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nO3dfXRc1Znn++8jqVSl0rtkyW8y2AmEBOiE2AqBJNDphgukhzbhNgxOd9+QNFnuSSeTdN8bBri5K02SxWqYMCHJ7dtMM4HhJSTEQ8IEwjDBEDKhVweInUB4cQAnvAkbW9b7S5Wkkp77x9klVcnSsaskY9n8PmuddfbZ5+xdu+zSfs7ZZ9cpc3dERETmU3G4GyAiIkubAoWIiMRSoBARkVgKFCIiEkuBQkREYlUd7gYstmXLlvnatWsPdzNERI4o27dv3+fubXPtO+oCxdq1a9m2bdvhboaIyBHFzF6Zb5+GnkREJJYChYiIxFKgEBGRWEfdPQoRkXJNTEzQ1dVFNps93E05ZFKpFB0dHSQSiYMuo0AhIhJ0dXVRX1/P2rVrMbPD3ZxF5+709PTQ1dXFunXrDrqchp5ERIJsNktra+tRGSQAzIzW1taSr5gUKEREChytQSKvnPenQCEiIrEUKERElrCrr76a66+//rC2QYFCRERiKVCIiCwx11xzDSeccAJnn302zz//PAC//OUvefe7383pp5/O5ZdfzsknnwzAyy+/zBlnnMH69etZv349//qv/7ro7dH0WBGROXz5vmd5btfgotZ54qoG/v5PT4o9Zvv27dx11138+te/JpfLsX79ejZs2MAnP/lJbrrpJj7wgQ9w5ZVXTh/f3t7O1q1bSaVSvPjii3zsYx9b9OfdKVCIiCwhjz76KBdeeCHpdBqAjRs3AjA0NMQHPvABAP78z/+cH//4x0D0JcHPfvazPPnkk1RWVvLCCy8sepsUKERE5nCgM/9DafYUVnef99gbbriB5cuX89RTTzE1NUUqlVr09ugehYjIEnLmmWdyzz33kMlkGBoa4r777sPMqK+v57HHHgPgrrvumj5+YGCAlStXUlFRwR133MHk5OSit0mBQkRkCVm/fj2XXHIJp5xyCn/2Z3/GGWecAcDNN9/M5s2bOf3003F3GhsbAfibv/kbbrvtNk477TReeOEFamtrF71NFndJcyTq7Ox0/XCRiJRjx44dvOtd7zrczZjT8PAwdXV1AFx77bXs3r2bb37zm2XVNdf7NLPt7t451/G6RyEicgS4//77+Yd/+AdyuRzHHnsst95665v22gccejKzW8xsr5k9U5D3NTP7rZn9xszuMbOmgn1XmdlOM3vezM4tyN9gZk+Hfd+ycLfGzJJm9v2Q/7iZrS0oc6mZvRiWSxfrTYuIHGkuueQSnnzySZ555hnuv/9+2trm/HnrQ+Jg7lHcCpw3K28rcLK7vxt4AbgKwMxOBDYBJ4Uy/2RmlaHMjcBm4Piw5Ou8DOhz9+OAG4DrQl0twN8D7wdOBf7ezJpLf4siIrIQBwwU7v5zoHdW3oPungubjwEdIX0BcJe7j7n7S8BO4FQzWwk0uPsvPLopcjvw0YIyt4X03cBZ4WrjXGCru/e6ex9RcJodsERE5BBbjFlPfwU8ENKrgdcK9nWFvNUhPTu/qEwIPgNAa0xd+zGzzWa2zcy2dXd3L+jNiIhIsQUFCjP7IpAD7sxnzXGYx+SXW6Y40/0md+909843c9xOROStoOxAEW4unw/8hc/Mse0C1hQc1gHsCvkdc+QXlTGzKqCRaKhrvrpERI5a+Smw5fjSl77EQw89BMA3vvENRkdHF6VNZQUKMzsPuALY6O6FLbkX2BRmMq0jumn9hLvvBobM7LRw/+HjwI8KyuRnNF0E/DQEnp8A55hZc7iJfU7IExGROXzlK1/h7LPPBhY3UBzwexRm9j3gw8AyM+simol0FZAEtoZZro+5+79z92fNbAvwHNGQ1GfcPf998k8TzaCqIbqnkb+vcTNwh5ntJLqS2ATg7r1m9lXgl+G4r7h70U11EZGj2de+9jW2bNnC2NgYF154IV/+8pcB+OpXv8qdd97JmjVrWLZsGRs2bOALX/gCn/jEJzj//PPZtWsXu3bt4o/+6I9YtmwZjzzyyILaccBA4e4fmyP75pjjrwGumSN/G3DyHPlZ4OJ56roFuOVAbRQRWXQPXAlvPL24da74A/jItQd16IMPPsiLL77IE088gbuzceNGfv7zn5NOp/nBD36w32PIC33uc5/j61//Oo888gjLli1bcLP1zWwRkSXowQcf5MEHH+S9730vED3C48UXX2RoaIgLLriAmpoaAP70T//0kLdFgUJEZC4HeeZ/qLg7V111FX/9139dlH/DDTe86W3R02NFRJagc889l1tuuYXh4WEAXn/9dfbu3cuHPvQh7rvvPrLZLMPDw9x///1zlq+vr2doaGhR2qIrChGRJeicc85hx44dnH766UA0bfY73/kO73vf+9i4cSPvec97OPbYY+ns7Jx+5HihzZs385GPfISVK1cu+Ga2HjMuIhIs5ceMF8o/cnx0dJQzzzyTm266ifXr1x90eT1mXETkKLd582aee+45stksl156aUlBohwKFCIiR5jvfve7b+rr6Wa2iIjEUqAQEZFYChQiIhJLgUJERGIpUIiILGFXX301119/fdEjxB999FFOOukkTjnlFDKZDJdffjknnXQSl19++SFpg2Y9iYgcAb7yla9Mp++8806+8IUv8MlPfhKAf/7nf6a7u5tkMnlIXluBQkRkibnmmmu4/fbbWbNmDW1tbWzYsGH6EeL9/f1s2bKFn/zkJzz00EMMDQ0xMjLC+9//fq666iouueSSRW+PAoWIyByue+I6ftv720Wt850t7+SKU6+IPWb79u3cdddd8z5G/FOf+hT/8i//wvnnn89FF10ERI/3ePLJJxe1rYUUKERElpBHH32UCy+8kHQ6DcDGjRsPc4sUKERE5nSgM/9DKfxy6JKhWU8iIkvImWeeyT333EMmk2FoaIj77rvvcDdJVxQiIkvJ+vXrueSSSzjllFM49thjOeOMMw53k/SYcRGRvCPlMeMLVepjxjX0JCIisRQoREQklgKFiEiBo204frZy3t8BA4WZ3WJme83smYK8FjPbamYvhnVzwb6rzGynmT1vZucW5G8ws6fDvm9ZmP9lZkkz+37If9zM1haUuTS8xotmdmnJ705EpASpVIqenp6jNli4Oz09PaRSqZLKHcysp1uBfwRuL8i7EnjY3a81syvD9hVmdiKwCTgJWAU8ZGbvcPdJ4EZgM/AY8D+A84AHgMuAPnc/zsw2AdcBl5hZC/D3QCfgwHYzu9fd+0p6hyIiB6mjo4Ouri66u7sPd1MOmVQqRUdHR0llDhgo3P3nhWf5wQXAh0P6NuBnwBUh/y53HwNeMrOdwKlm9jLQ4O6/ADCz24GPEgWKC4CrQ113A/8YrjbOBba6e28os5UouHyvpHcoInKQEokE69atO9zNWHLKvUex3N13A4R1e8hfDbxWcFxXyFsd0rPzi8q4ew4YAFpj6tqPmW02s21mtu1oPhMQETkcFvtm9lzfO/eY/HLLFGe63+Tune7e2dbWdlANFRGRg1NuoNhjZisBwnpvyO8C1hQc1wHsCvkdc+QXlTGzKqAR6I2pS0RE3kTlBop7gfwspEuBHxXkbwozmdYBxwNPhOGpITM7Ldx/+PisMvm6LgJ+6tGUg58A55hZc5hVdU7IExGRN9EBb2ab2feIblwvM7MuoplI1wJbzOwy4FXgYgB3f9bMtgDPATngM2HGE8CniWZQ1RDdxH4g5N8M3BFufPcSzZrC3XvN7KvAL8NxX8nf2BYRkTePnvUkIiJ61pOIiJRPgUJERGIpUIiISCwFChERiaVAISIisRQoREQklgKFiIjEUqAQEZFYChQiIhJLgUJERGIpUIiISCwFChERiaVAISIisRQoREQklgKFiIjEUqAQEZFYChQiIhJLgUJERGIpUIiISCwFChERiaVAISIisRQoREQklgKFiIjEUqAQEZFYCwoUZvZ3ZvasmT1jZt8zs5SZtZjZVjN7MaybC46/ysx2mtnzZnZuQf4GM3s67PuWmVnIT5rZ90P+42a2diHtFRGR0pUdKMxsNfA5oNPdTwYqgU3AlcDD7n488HDYxsxODPtPAs4D/snMKkN1NwKbgePDcl7Ivwzoc/fjgBuA68ptr4iIlGehQ09VQI2ZVQFpYBdwAXBb2H8b8NGQvgC4y93H3P0lYCdwqpmtBBrc/Rfu7sDts8rk67obOCt/tSEiIm+OsgOFu78OXA+8CuwGBtz9QWC5u+8Ox+wG2kOR1cBrBVV0hbzVIT07v6iMu+eAAaB1dlvMbLOZbTOzbd3d3eW+JRERmcNChp6aic741wGrgFoz+8u4InPkeUx+XJniDPeb3L3T3Tvb2triGy4iIiVZyNDT2cBL7t7t7hPAD4EPAHvCcBJhvTcc3wWsKSjfQTRU1RXSs/OLyoThrUagdwFtFhGREi0kULwKnGZm6XDf4CxgB3AvcGk45lLgRyF9L7ApzGRaR3TT+okwPDVkZqeFej4+q0y+rouAn4b7GCIi8iapKreguz9uZncDvwJywK+Bm4A6YIuZXUYUTC4Oxz9rZluA58Lxn3H3yVDdp4FbgRrggbAA3AzcYWY7ia4kNpXbXhERKY8dbSfonZ2dvm3btsPdDBGRI4qZbXf3zrn26ZvZIiISS4FCRERiKVCIiEgsBQoREYmlQCEiIrEUKEREJJYChYiIxFKgEBGRWAoUIiISS4FCRERiKVCIiEgsBQoREYmlQCEiIrEUKEREJJYChYiIxFKgEBGRWAoUIiISS4FCRERiKVCIiEgsBQoREYmlQCEiIrEUKEREJJYChYiIxFpQoDCzJjO728x+a2Y7zOx0M2sxs61m9mJYNxccf5WZ7TSz583s3IL8DWb2dNj3LTOzkJ80s++H/MfNbO1C2isiIqVb6BXFN4H/6e7vBN4D7ACuBB529+OBh8M2ZnYisAk4CTgP+Cczqwz13AhsBo4Py3kh/zKgz92PA24Arltge0VEpERlBwozawDOBG4GcPdxd+8HLgBuC4fdBnw0pC8A7nL3MXd/CdgJnGpmK4EGd/+Fuztw+6wy+bruBs7KX22IiMibYyFXFG8DuoH/ama/NrNvm1ktsNzddwOEdXs4fjXwWkH5rpC3OqRn5xeVcfccMAC0zm6ImW02s21mtq27u3sBb0lERGZbSKCoAtYDN7r7e4ERwjDTPOa6EvCY/LgyxRnuN7l7p7t3trW1xbdaRERKspBA0QV0ufvjYftuosCxJwwnEdZ7C45fU1C+A9gV8jvmyC8qY2ZVQCPQu4A2i4hIicoOFO7+BvCamZ0Qss4CngPuBS4NeZcCPwrpe4FNYSbTOqKb1k+E4akhMzst3H/4+Kwy+bouAn4a7mOIiMibpGqB5f89cKeZVQO/Bz5JFHy2mNllwKvAxQDu/qyZbSEKJjngM+4+Ger5NHArUAM8EBaIbpTfYWY7ia4kNi2wvSIiUiI72k7QOzs7fdu2bYe7GSIiRxQz2+7unXPt0zezRUQklgKFiIjEUqAQEZFYChQiIhJLgUJERGIpUIiISCwFChERiaVAISIisRQoREQklgKFiIjEUqAQEZFYChQiIhJLgUJERGIpUIiISCwFChERiaVAISIisRQoREQklgKFiIjEUqAQEZFYChQiIhJLgUJERGIpUIiISCwFChERiaVAISIisRYcKMys0sx+bWY/DtstZrbVzF4M6+aCY68ys51m9ryZnVuQv8HMng77vmVmFvKTZvb9kP+4ma1daHtFRKQ0i3FF8XlgR8H2lcDD7n488HDYxsxOBDYBJwHnAf9kZpWhzI3AZuD4sJwX8i8D+tz9OOAG4LpFaK+IiJRgQYHCzDqAfwN8uyD7AuC2kL4N+GhB/l3uPubuLwE7gVPNbCXQ4O6/cHcHbp9VJl/X3cBZ+asNERF5cyz0iuIbwH8Apgrylrv7boCwbg/5q4HXCo7rCnmrQ3p2flEZd88BA0Dr7EaY2WYz22Zm27q7uxf4lkREpFDZgcLMzgf2uvv2gy0yR57H5MeVKc5wv8ndO929s62t7SCbIyIiB6NqAWU/CGw0sz8BUkCDmX0H2GNmK919dxhW2huO7wLWFJTvAHaF/I458gvLdJlZFdAI9C6gzSIiUqKyryjc/Sp373D3tUQ3qX/q7n8J3AtcGg67FPhRSN8LbAozmdYR3bR+IgxPDZnZaeH+w8dnlcnXdVF4jf2uKERE5NBZyBXFfK4FtpjZZcCrwMUA7v6smW0BngNywGfcfTKU+TRwK1ADPBAWgJuBO8xsJ9GVxKZD0F4REYlhR9sJemdnp2/btu1wN0NE5IhiZtvdvXOuffpmtoiIxFKgEBGRWAoUIiISS4FCRERiKVCIiEgsBQoREYmlQCEiIrEUKEREJJYChYiIxFKgEBGRWAoUIiISS4FCRERiKVCIiEgsBQoREYmlQCEiIrEUKEREJJYChYiIxFKgEBGRWAoUIiISS4FCRERiKVCIiEgsBQoREYmlQCEiIrHKDhRmtsbMHjGzHWb2rJl9PuS3mNlWM3sxrJsLylxlZjvN7HkzO7cgf4OZPR32fcvMLOQnzez7If9xM1tb/lsVETk65Can6Bke4/fdw/zq1T4eeX4vfSPjh+z1qhZQNgf8X+7+KzOrB7ab2VbgE8DD7n6tmV0JXAlcYWYnApuAk4BVwENm9g53nwRuBDYDjwH/AzgPeAC4DOhz9+PMbBNwHXDJAtosIrIkuDuj45P0jY4zkJlgYHSC/swEA5kJ+kej9UBmvGg7vx4ey+1X352fej8fPG7ZIWlr2YHC3XcDu0N6yMx2AKuBC4APh8NuA34GXBHy73L3MeAlM9sJnGpmLwMN7v4LADO7HfgoUaC4ALg61HU38I9mZu7u5bZbRGSxjeem6B8dp290Yv91Zpz+kQn6Rsfpz8zkD4xOMD45NW+d1ZUVNKYTNNYkaKpJsKIhxQkr6sN2NY01VTSlq6ePOb697pC9v4VcUUwLQ0LvBR4HlocggrvvNrP2cNhqoiuGvK6QNxHSs/PzZV4LdeXMbABoBfbNev3NRFckHHPMMYvxlkTkLWhqyhnK5ugbHY869tGJkI46+Pz27PXo+OS8dVZXVtCUTtCcrqYpnWDdslrWp6tpCtvN6QSNNVG6sSYxva5JVBJG4Q+7BQcKM6sDfgD8rbsPxryxuXZ4TH5cmeIM95uAmwA6Ozt1tSEi051+z8gYfaPj9I5M0DcyTu/oOL0j0dI3UhwQBjITTM3Tg5hBU01iuoNfHs7wm9PVUWcf1s3pahprEjTXRttLqcMv14IChZkliILEne7+w5C9x8xWhquJlcDekN8FrCko3gHsCvkdc+QXlukysyqgEehdSJtF5MiTH8/vDR37zLqg8x+O1vnOv290gsl5ev3qqgpaa6ujTr42wcqmGprT0ZBO/uy/uTYKCvlA0JBKUFFxZHf45So7UISZSTcDO9z96wW77gUuBa4N6x8V5H/XzL5OdDP7eOAJd580syEzO41o6OrjwP87q65fABcBP9X9CZEjX3ZicqbDH5mY7uDjAsF4bu7x/AqDlulOv5q3t9XRUldNS9huqY06/tbaJM21CVpqq4+Ks/w300KuKD4I/B/A02b2ZMj7v4kCxBYzuwx4FbgYwN2fNbMtwHNEM6Y+E2Y8AXwauBWoIbqJ/UDIvxm4I9z47iWaNSUiS8zUlDOQmaBnZIx9w+P0DI+zb3iMnuEx9o2M0zM8VpA3ztAcs3bymtKJ6U5+dVMNf7C6IQzjFHf+TelqWmur39Jn+m8WO9pO0Ds7O33btm2HuxkiR7Tc5BT9mYlZZ/kT02f7PcNj9IyMh6AwRu/IOLk5hnnyZ/uttUla66pZVhetW2uraa1LRp1/wVl/Y02Cqkp9D/hwMLPt7t45175FmfUkIkuXuzM8lqM3dOz5oZyi9XRAmKB3JLqpO5/a6kqaa6NOf3VTinevbmRZ/UwwaKtL0hoCQnO6mkqd7R/xFChEjjDuzmA2Fzr3aCinZyR/ph/yptPRMt98/erKimh8P5zVr2qqmR7vz+e3Fmw3pROkEpVv8juWw02BQmSJyE5M0j00xt6hMbqHstPpvYNjdA+PsTfk9Y6MMzE595BxbXVldCO3NsmKxhQnrmqYHuppqU2G9cySrtZNXTkwBQqRQ8jdGRrLsXcwy57BMfYMZguCwUznv3dojKHs/jd4Kwxa65K01SVpb0jyrhUN0bBO6OijIJCkJQQDne3LoaBAIVKmzPgkewaz0TI0xp6BgvRgdjo4ZCb2/9ZuTaKS9oYoAJywop4PHbeM9oYUbXVJ2hpmAkNrbVJj/HLYKVCIzDIylgtDPtlonT/zHxxjz9DMlcFcVwCpRAXLG1Isb0jxBx1NnF2fZHlDivaGJO31KZY3JGlvSFGrIR9ZJJNTk4zkRqiprCFRmTgkr6FAIW8J7k7/6MR0p793sCAAFA4FDWYZmeO5PYlKo60uyfLGFMe11fHBt7fS3pBiRQgK+QDQkKpSAJCDMuVTjEyMMDQ+xND4ECMTIwxPDM+sx0cYmgj54zP5w+PDRcdlchkAbjn3Ft634n2HpK0KFHJEKwwA+WGg/NXAnnAFsHcwCgJzzfypra6krT462z9xVQMfPqGN9voU7fXR0E9+X1ONvtQlxSanJhmeGJ7u6IfGhxiaGCreHh9icHywaHt4YpjB8UGGx4fx/R9dV8Qw6hJ11FXXUZuopS5RR2OqkdX1q6P8RB211VF+R11HbF0LoUAhS9bIWI7dAxl2DxTfCC4OCHMHgIZU1fSQz/vXtdAWhn7a65MhCETp2qT+BN6qJqcm5+3gZ3fucx0zPDF8wNeoS9RRX10/vaysW0lDdcNMXiJa5wNBfaJ+uuOvS9RRU1WzJK5Q9Vcih8XoeI5d/VneGMiyayDD7v4sbwxmivLmugdQnw8A9Unet7aF9oYky+ujgLC8ITWd1uyft478mX3/WD8DYwPT68Hxwel00TIeHTM0PnTAuvMdeX5ZXbe6aDu/v6jzD0tdoo7KiqPjc6hAIYsuOzHJrv7oSiC/jpZMFAT6MwzOEQSW1VWzojHFMa1pTntbCysaa1jVFN0HWNGYor0+RU310fGHV4rcVI7B8UEGxgYYGh8iVZWa7pjSVek39YzT3cnkMozmRsnkMmRyGbK57JzpTC5DdjJLZiKsC/NzWXJTORzH3XGcKZ8q2nZ3ppiKtvN5BftzU7nps/u4IZz66noaqxtpSjbRmGxkTcOa6XRDdUNRJ99Q3UBddXQVUFtVe9R09AulQCElmZicYs9gtigI7OqPrgTyw0S9c/x2b2ttFAQ6mtOcuq6FFY0pVjXWTK+P9quAKZ8im8syPDFcdGY7ODZYnB6fOfPNB4e4IY4Kq4g9s62rrpvJC8ekqlKMToxO3wwdnRidvjk635I/biQ3wpTP/6tssxlGqipFTVXN9JKqTJGqihbDqLAKMKigAjPDsJl12L9fvkX59Yl6mlJNNFY30pgsWMJ2Q3XDkdfZu8PUJPgkTOWi9FQOfGpme3pfyPNJaDoWkofmV+4UKGSau9M7Ms6u/iyv92dCAAjBIAwP7R3K7vfDLvWpKlaFs//3rGliVWOKlY01rGyaCQZHShBwd0YmRoo67HwnPjwxzFhujOxklmwuy9jkTDo7mY325dOTY0X541PxP3xfZVU0JBumO7n2dDvHNx9PQ3VDUedXV11HNpfdfyy9YPz85cGXp9OjudGDfu8VVjF9w7Q2URulq+tYXrt8v/x0VZqaxEzHXxQICgJDsjK58CueyQnI9MFoL2R6Z9LZAcjlYGgM/A3w3VFn6lOAz6SLFg9LfnsydLz5dGFewb6i7cIyc+wr6ugnC+qdq26fSec7/BICcZGP3wtv+8OF/VvPQ4HiLSQ7MTk99PN6uArY1Z9h10BmOjBkJ4o/pKlEBatCp/+h45exqjHFqqYaVjbVRAGhqYa6JXpDeHJqkr6xPvaO7mVfZl/RWXp+PfvMfnB8kMnpp9/PLVGRmD4rTlYmo7PjyhTJqiSNyUaWVy3fLz9VGR1bX11PQ7IhGvooOAs+VENIE1MT0TTL8SEGJ6Kgks1lo84+kS7q/FOVqUMzjOUOE6MwPlK8ZAeijn80dP6z05k+GO2Dg7iXMCer2H/BCrZDuqIybFfOpCsqC7YroaJi1nZ+XQVVyTn22Rx5FXPXnc+rCPXl9xW+xvS+ilnHVUVtq6iC9hMX9b+t0NL8C5eyDGQmeL0vQ1ffKF19maKrgtf7s+wbHtuvTHt9klVNNbxrRQNnvbOdVU01rGqqYXVYmtKJJTHrolBhAOge7aY701203pvZy77RfezL7pt3mKQ+UT99Bt9Q3cDK2pVFHXdDdUO0vyCvLlFHsjJ5RA1lJCoSNKWaaEo1RRnuMD4MuXGYHIPcGIz1wuRumBwP+QVLbiw6o58cK96fy0b1jOcDwPDcwWB8JMo/wDRQMKhpjpZ0C9Qth7Z3Rel8fn5fTT6vCSoScwQEixZZNAoURwj36IdhuvoyYZkJBvnt2bOEahKVrG6OOv4TVzWE4aGZQLC8MUmyanE6vYnJCfaM7uGNkTeK1ntG9jA2OTY9Bp0fby5MF45DV1jF9Fh1hVVg4WfT+7J90wGgJ9sz51l/S6qFtpo2lqWXcULzCSyrWUZ7up22mjZaa1qnb2DWV9dTVXGUf/QzfdDze+jZCb2/g57fhfXvYWxgcV6jqgaq01BdC9V10TqRhvSykJcuzp8+Lg2J2qijzweAVFN0ZixL0lH+13JkGcxO8FrvaFj2DwbDs34VrLa6ko7mNB3NNZy6tpmO5jSrm2voaK6hozlN8yJdDUxMTdA92l0UBGane7I9+5WrT9SzvHY56ar09KyW/MyW6fSsGS+F+dMzXnCak837B4B0G201bbSn22lNtR6yxxcsWdnBgiDw+4Jg8Lto6GaaQdMaaHk7vPtiaFwDiRqorI6GTSqrZ5aqfDoJlYn591eloiEQeUtQoHgTjeemeMZKMfoAAA0FSURBVL0/w2u9o7zaO8prfTNB4dXe0f1+LKY+WRU6/jSnva01BICa6eDQWFN6IBibHCueaz5rpk0+nc/vyfSwL7Nvv+mHtYlaVqRXsKJ2BSe0nDCdXp5eHq3DDdC3vPz4fKY/GpOfyOw/jDMZhndyY7OGfQr259MTGeh/JQoGI3uLX6thNbS8DU7cGAWF1uOg9e3QvDbq8EXKpECxiNyd7uGx6UDwak+mIBiMsnswS+Evz1ZXVkQdf0ua96xpZE1zmjUtaY5pSbOmOU1jev4z5NxUjt5s73Snn+/4+8b69v+SUUHnn53MzltnlVUVTTFckV7BO1veyYraFaxIR51/PiDUVR+aaXhLivvM+HxuHCZGQoffP9PxT6dnrwv2Tc3/a3EHZJXhrD4RneVXpaKrg3ecGwWBlreHYLAuGtIROQQUKEo0MTnFrv4Mr/SM8krvKK/2jPBKTwgMvaOMznqg3PKGJGvyVwQhCKxsrKKt0ahPTZGZCl9Cmsgwmusjk8uwM5PhN4Oj0x38dCDIzlwJDE3MPxMkUZGYHo9vTDaypm4NJ7eeXHSjdvZ880M586YkE1kYfB0GXoP+12CgK9qe2v8LeiXxqZkz9qL1WPGN3dnHHPAmLNEN1FQTpBqjcfdUU9SZz85LNUbj9NPDOMkDD/loeEeWAAWKOQyP5Xi1Z5RXe0cKAsIor/SOsKs/y+TUJFRmqKgcIVGdoa1pnOamcf5gVZZkdQarHMYrsmDjZCejQPBUbpTH9mXI7M6Q84Pv9OoSdTQmo2+VNiWbOKbhmOl0YX5Tqmk6vVSeD7Mf9+gma2EQGHiteHv2cAoGde2LMHRiofNNhk45GZ2BVzZH+bP37bdORjdkCzv9fDpZr1k2clR7ywaK4bEcL+8b4eWeEV7eN8JL+4b5Xc8bvDr0OoMT+7DKEaxqGKscIZnMkEyOUrlilJaVw4xNDeLMTLscCAsZqJ+spyXVQn2inpqqGtqT7dNfPkon0kVfTEpXpefdV1NVQ0OygUTFIt6gncxFwyfTUxeHD5zOZZkZL/Oi1cz27P1enB7tCUGhK3r9QlWp6OZqYwcsPwmajonS+byG1VFnLSKHzVEdKAqDwe+6B3h+Xxcv93exe2Q3I1PdWKKfikQ/FYk+KhL9kJ6ENNQU1FGXqKe1poWWVAvNyQ6aU820pML2rHRzsnnxZt64R530aH/ouIdhLN+BD82TLjxueO5O/2BVVEVTGatSBWfLVpyGObbZf3+qCdreAcedVRwEmo6BdKvOxkWWuKMuULzc28PZ/+Va9mX3kPV9WAgCVjWImUMKSEESqK9qYXntCo5t2MAxDatZWbeSVbWrWF67PASGRez4IRp/H90Hw3thZB+MdBcs+6Jhl+n0voO/CVqRiJ7xUh2WZJi7Xrd8Zh574Vz32HTY1lm8iARHRKAws/OAbwKVwLfd/dr5jh3xN9hTfSdWXUlL1TLaUitYU38ib2/p4NjGDlbWrmRV3SpW1K4gWVniuHd+quPYMIwNwdhgOIMf2n/J9O0fCMYG5663qgbq2qC2LRpqWfme6EtLqcZo/LuwM0/W759Wpy4ih9CSDxRmVgn8f8D/BnQBvzSze939ubmOX1fTztZTrqCtIkXl1GR0Vj45Ho3Pd++CPa+EKY8hfyoX9oe8qYlorvp0ACgIBONDB/fALquMHjVQ2wa1y2DVe6G2PUrXthUsy6IbtdX6voGILF1LPlAApwI73f33AGZ2F3ABMGegSPd3seKezxxczfmpiRVVYZ56SFelorP1ZD3Ur4Bkw8zZe7I+GtrZL69gKRrXFxE5sh0JgWI18FrBdhfw/sIDzGwzsDlsjtmXB59Z4GsuA/YdBXUshTYslTqWQhuWSh1LoQ1LpY6l0IalUscJ8+04EgLFXKfmRd+CcvebgJsAzGybu3cu6AWPkjqWQhuWSh1LoQ1LpY6l0IalUsdSaMNSqcPMts2370h4XGMXsKZguwPYdZjaIiLylnMkBIpfAseb2TozqwY2Afce5jaJiLxlLPmhJ3fPmdlngZ8QTY+9xd2fjSly0yK87NFSx1Jow1KpYym0YanUsRTasFTqWAptWCp1zFve3A/ioWciIvKWdSQMPYmIyGGkQCEiIrGOmkBhZreY2V4zK/s7FGa2xsweMbMdZvasmX2+xPIpM3vCzJ4K5b+8gLZUmtmvzezHZZZ/2cyeNrMn46a9HaCOJjO728x+G/5NTi+x/Anh9fPLoJn9bYl1/F34t3zGzL5nZqnS3gWY2edD+WcP9vXn+jyZWYuZbTWzF8O6uYw6Lg7tmDKz2KmM85T/Wvj/+I2Z3WNmTWXU8dVQ/kkze9DMVpVaR8G+L5iZm9myMtpxtZm9XvD5+JNS22Bm/97Mng//pv+xjDZ8v+D1XzazJ8uo4xQzeyz/t2Zmp5ZRx3vM7Bfhb/Y+M2uIKT9nP1Xq57Mk7n5ULMCZwHrgmQXUsRJYH9L1wAvAiSWUN6AupBPA48BpZbbl/wS+C/y4zPIvA8sW+G96G/CpkK4GmhZQVyXwBnBsCWVWAy8BNWF7C/CJEl/3ZOAZIE00eeMh4PhyPk/AfwSuDOkrgevKqONdRF9s+hnQWUb5c4CqkL6uzDY0FKQ/B/znUusI+WuIJpm8cqDP2jztuBr4wkH+P85V/o/C/2cybLeX8z4K9v8n4EtltONB4CMh/SfAz8qo45fAH4b0XwFfjSk/Zz9V6uezlOWouaJw958DvQc8ML6O3e7+q5AeAnYQdVYHW97dfThsJsJS8mwBM+sA/g3w7VLLLpZwRnMmcDOAu4+7e/8CqjwL+J27v1JiuSqgxsyqiDr7Ur9D8y7gMXcfdfcc8L+ACw9UaJ7P0wVEwZOw/mipdbj7Dnd//mAaPk/5B8P7AHiM6HtFpdZR+HTKWg7wGY3527oB+A8HKn+AOg7KPOU/DVzr7mPhmNm/enXQbTAzA/4t8L0y6nAgfwXQyAE+o/PUcQLw85DeCvxZTPn5+qmSPp+lOGoCxWIzs7XAe4muCkopVxkuX/cCW929pPLBN4j+AA/iCYTzcuBBM9tu0SNOSvU2oBv4r2EI7NtmtpCnF27iAH+Es7n768D1wKvAbmDA3R8s8XWfAc40s1YzSxOd8a05QJn5LHf33aFtu4H2MutZLH8FPFBOQTO7xsxeA/4C+FIZ5TcCr7v7U+W8foHPhmGwW8oYKnkHcIaZPW5m/8vM3reAdpwB7HH3F8so+7fA18K/5/XAVWXU8QywMaQv5iA/o7P6qUP2+VSgmIOZ1QE/AP521tnXAbn7pLufQnSmd6qZnVzia58P7HX37aWUm8MH3X098BHgM2Z2Zonlq4guj2909/cCI0SXsyWz6IuSG4H/VmK5ZqKzpHXAKqDWzP6ylDrcfQfREM1W4H8CTwEL/AHuw8/Mvkj0Pu4sp7y7f9Hd14Tyny3xtdPAFykjwMxyI/B24BSiE4H/VGL5KqAZOA24HNgSrgzK8TFKPJEp8Gng78K/598RrsJL9FdEf6fbiYaTxg9UYCH9VKkUKGYxswTRP/6d7v7DcusJwzQ/A84rsegHgY1m9jJwF/DHZvadMl5/V1jvBe4hegpvKbqAroIroruJAkc5PgL8yt33lFjubOAld+929wngh8AHSn1xd7/Z3de7+5lEl/zlnDUC7DGzlQBhHTvUcaiY2aXA+cBfeBiQXoDvEjPMMY+3EwXvp8LntAP4lZmtKKUSd98TTqymgP9CeZ/RH4Yh3yeIrsBjb6rPJQxr/u/A90stG1xK9NmE6GSo1PeBu//W3c9x9w1EAet3ccfP008dss+nAkWBcDZyM7DD3b9eRvm2/CwUM6sh6uh+W0od7n6Vu3e4+1qi4ZqfuntJZ9FmVmtm9fk00Q3QkmaDufsbwGtmln+i5FnM82j3g1Du2dqrwGlmlg7/N2cRjceWxMzaw/oYog6h3DPHe4k6BcL6R2XWUzaLfsTrCmCju4+WWcfxBZsbKf0z+rS7t7v72vA57SK6ufpGie1YWbB5ISV+RoH/DvxxqOsdRBMuynl66tnAb929q4yyEN2T+MOQ/mPKOBEp+IxWAP8P8J9jjp2vnzp0n8/Fuit+uBeiP/7dwATRB/eyMur4ENHY/m+AJ8PyJyWUfzfw61D+GQ4wg+Ig6vswZcx6Irq/8FRYngW+WObrnwJsC+/nvwPNZdSRBnqAxjLb8GWijuwZ4A7CDJcS63iUKMg9BZxV7ucJaAUeJuoIHgZayqjjwpAeA/YAPymx/E6ix+7nP58HmrE0Vx0/CP+evwHuA1aXWses/S9z4FlPc7XjDuDp0I57gZUllq8GvhPey6+APy7nfQC3Av9uAZ+LDwHbw+frcWBDGXV8nmj20gvAtYSnZsxTfs5+qtTPZymLHuEhIiKxNPQkIiKxFChERCSWAoWIiMRSoBARkVgKFCIiEkuBQkREYilQiIhIrP8fiKNPjcVy80cAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "dga_pair_matches_df.plot(legend=True, ylim=(0,len(data_df)), xticks=range(0,21))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 136,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 265
    },
    "colab_type": "code",
    "id": "GKSH4nBB6R2A",
    "outputId": "a14ee1b5-7030-4197-c897-310489ba6e2b"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAD4CAYAAADsKpHdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nOydd1hU19aH3z3D0HtH6VXBgtg7RhFMosaYRONNLxqTaG7aTbz50qMx1dSbxBRLmjWxRppibwEVFRRRiqAIVgTpcL4/DhhUVIQZZoDzPs88M+zZZ+81OjO/2WuvvZaQJAkFBQUFBQWVvg1QUFBQUDAMFEFQUFBQUAAUQVBQUFBQqEURBAUFBQUFQBEEBQUFBYVajPRtwM1wdHSUvL299W2GgoKCQqsiKSnpjCRJTrdyjcELgre3N4mJifo2Q0FBQaFVIYTIvtVrFJeRgoKCggKgCIKCgoKCQi2KICgoKCgoAK1gD0FBQUGhKVRWVpKbm0tZWZm+TdEppqamuLu7o9Fomj2WIggKCgptktzcXKysrPD29kYIoW9zdIIkSZw9e5bc3Fx8fHyaPZ7iMlJQUGiTlJWV4eDg0GbFAEAIgYODg9ZWQYogKCgotFnashjUoc3XaPCCUFhayemicn2boaCgoNDmMfg9hOPnSug9Mx4/Jwv6+TrQ19eBfj72OFub6ts0BQUFhVvirbfewtLSkpdeeknfpjSIwQuCn5Ml00d1YlfGWVbuO8mvu44D4OtoQV9fe/r6ONDX1x43GzM9W6qgoKDQujF4QTA3VvPUUD+eGupHVXUNqXkX2ZVxjp0ZZ1mzP4/fd+cA4OVgTl8f+8uriI62ikAoKCjon5kzZ7Jw4UI8PDxwcnKiZ8+e/P333zz++ONYWFgwaNAg1q1bx8GDB8nKyuLBBx/k0qVLAHz11VcMGDCgxWw1eEGoj5FaRTd3W7q52/LkEF+qayQO5V1kZ8ZZdmWeIyYlnyWJuQC425nJ4uBjzwB/R0UgFBTaMW+vTiH15EWtjhncwZo3R4fcsE9SUhKLFi1i7969VFVVERYWRs+ePXn00UeZO3cuAwYM4NVXX73c39nZmbi4OExNTUlPT+f+++9v0VxurUoQrkatEnTpaEOXjjY8MdiXmhqJw6eK2JV5ll0Z51h/KJ9lSbkIASODXZgy1I8wTzt9m62goNBO2LJlC+PGjcPc3ByAMWPGAFBUVHT5l/+kSZNYs2YNIB+me/bZZ9m3bx9qtZojR460qL2tWhCuRqUSBHewJriDNY8O9KGmRiK9oJg1+0+ycEc2MSn59PG2Z8pQX4YFOaNStf2QNAUFBW76S16XXB0WKknSdfvOmTMHFxcXkpOTqampwdS0ZYNnbhp2KoT4SQhRIIQ42MBzLwkhJCGEY722GUKIo0KINCFEZL32nkKIA7XPfSFaIEBYpRIEuVrx4sggtr96G2/cGcyJC6U8viCRyM82szQxh4qqGl2boaCg0E4ZMmQIf/75J6WlpRQVFbF69WqEEFhZWbFz504AFi1adLl/YWEhbm5uqFQqfv75Z6qrq1vU3sacQ5gPRF3dKITwACKA4/XagoGJQEjtNf8TQqhrn/4GmAwE1N6uGVOXWJgY8dggHza+HM5nE0JRqwQvL9vPkA8TmLv5GEVllS1pjoKCQjsgLCyMCRMmEBoayvjx4xk8eDAAP/74I5MnT6Z///5IkoSNjQ0ATz/9NAsWLKBfv34cOXIECwuLFrVX3Gj5crmTEN7AGkmSutRrWwa8C6wEekmSdEYIMQNAkqT3a/vEAG8BWUCCJEmdatvvB8IlSZpys7l79eol6WJTRZIkNqef4duNx9iRcRYrEyP+1c+LxwZ6K2ccFBTaAIcOHaJz5876NqNBiouLsbS0BGD27Nnk5eXx+eefN3m8hl6rECJJkqRetzJOk/YQhBBjgBOSJCVf5fnpCOys93dubVtl7eOr2/WGEIKhgU4MDXQiOecCczdnMHfzMX7amsndYR15cogvfk6W+jRRQUGhjbJ27Vref/99qqqq8PLyYv78+fo2CWiCIAghzIHXgJENPd1Am3SD9uvNMRnZvYSnp+etmnjLdPew5et/hZF15hI/bM1gaWIuixNziOgsRyb19FIikxQUFLTHhAkTmDBhgr7NuIam5DLyA3yAZCFEFuAO7BFCuCL/8veo19cdOFnb7t5Ae4NIkjRXkqRekiT1cnK6pRrRzcLb0YL37urKtldv49lh/uzKPMf4b7Zz77fbiU/Np6bm5u41BQUFhdbKLQuCJEkHJElyliTJW5Ikb+Qv+zBJkk4Bq4CJQggTIYQP8ubxbkmS8oAiIUS/2uiih5D3HgwSR0uTKyKTTl4o44mFcmTS8qRcKquVyCQFBYW2R2PCTn8HdgBBQohcIcTj1+srSVIKsARIBaKBZyRJqoubmgr8ABwFjgHrmmm7zqkfmTRnQndUQvDi0mSGfpjAvG2ZlFRU6dtEBQUFBa3RqCgjfaKrKKOmIEkSCWkFfLPxGH9nncfOXMMjA3x4eIAXtubG+jZPQUGhHoYcZaRttBVlZPD1EAwJIQS3dXJh6VMDWPpUf8I87ZgTf4QBszfw7ppU8gpL9W2igoKCAVEXWtoU3njjDeLj4wH47LPPKCkp0ZZZ10VZITSTw6cu8t2mDFYln0Ql4K7QjkwZ6oe/sxKyqqCgTwxhhWBpaUlxcXGzx/H29iYxMRFHR8cGn1dWCAZCJ1dr5kwIZeNL4Uzq48mq5JNEzNnElJ8T2ZdzQd/mKSgoGAgfffQRvXv3plu3brz55puX29999106depEREQE999/Px9//DEAjzzyCMuWLeOLL77g5MmTDBs2jGHDhunUxjaV3E6feNib8/bYLkwbHsCC7Vks2J5FTEo+A/wcmBruxyB/x3ZR31VBwSBZ9yqcOqDdMV27wqjZjeoaGxtLeno6u3fvRpIkxowZw+bNmzE3N2f58uXXpMeuz/Tp0/n0009JSEi47gpBWyiCoGXqQlanDPXj913H+WFrBg/+uJsuHa2ZOtSfqC6uqJUsqwoK7YrY2FhiY2Pp0aMHIKeuSE9Pp6ioiLFjx2JmJtdrGT16tD7NVARBV1iaGPHkEF8eGuDFyr0n+XbTMZ75bQ/eDuY8M8yfcT06YqRWPHYKCi1CI3/J6wpJkpgxYwZTplyZvm3OnDl6sqhhlG8kHWNipOa+3h7EvTCUbx8Iw9LUiJeX7WfEp5v4Y08u1crpZwWFNk9kZCQ//fTT5Q3mEydOUFBQwKBBg1i9ejVlZWUUFxezdu3aBq+3srKiqKhI53YqK4QWQq0SRHVxIzLElbjUfObEp/PCkmS+SjjKc8MDuLNbB8WVpKDQRhk5ciSHDh2if//+gBx99Msvv9C7d2/GjBlD9+7d8fLyolevXpdTYddn8uTJjBo1Cjc3NxISEnRmpxJ2qidqaiRiU0/xWXw6h08V4e9syXPDA7ijq5tSyU1BQQsYQthpY6hLhV1SUsKQIUOYO3cuYWFhtzSGEnbaylHVrhj+mj6YryeFIYBpv+9l1Odb+OtAnpJIT0GhnTB58mRCQ0MJCwtj/PjxtywG2kRxGekZlUpwRzc3orq4svZAHp/HH+HpX/fQydWKf48IJDLERQlXVVBow/z222/6NuEyygrBQFCrBGO6dyD2+aF8PjGUiqoanvoliTu+2Epcav4NC3MrKCgoaANFEAwMtUowNrQjsc8P4dP7unOpooonFyYy5qttbDisCIOCgoLuMHhBKK1qnwnjjNQq7g5zZ/0LQ/nonm5cKK3gsfmJ3PW/7SSkFSjCoKCgoHUMXhBOFJ+grKpM32boDSO1int7ebDhxXA+GN+Vs8XlPDrvb8Z+vY3fdh3nYlmlvk1UUFBoIxi8IJRXl/P5ns/1bYbe0ahVTOjtyYYXw5k1ritlldX8988D9JkZz78X7WX70TNKZJKCgoHz1ltv8fHHH1+R2nrLli2EhIQQGhpKaWkpL7/8MiEhIbz88sstbp/BRxnZm9rzy6FfGOYxjD5uffRtjt4xNlIxqa8n9/fxYH9uIUuTcli57yQr9p3E3c6M8WHu3NPTHQ97c32bqqCgcB3eeeedy49//fVXXnrpJR599FEAvvvuO06fPo2JiUmL23XTg2lCiJ+AO4ECSZK61LZ9BIwGKpDLYT4qSdKF2udmAI8D1cB0SZJiatt7AvMBM+Av4DmpEY7wsJ5hUsf/60hZdRl/jPkDK2OrJr3QtkxZZTUxKadYlpTL1qNnkCQY4OfAvb3ciQpxw8xYrW8TFRRaHEM5mDZz5kwWLlyIh4cHTk5O9OzZk4MHD3LnnXdy4cIF/vOf/2BjY8OAAQMoKipi7dq1dO3alRkzZjBhwoRGzaGtg2mNWSHMB74CFtZriwNmSJJUJYT4AJgBvCKECAYmAiFAByBeCBFYW1f5G2AysBNZEKJoRF1llVAxa9AsHlz3ILN3z2bmoJmNf3XtBFONmrGhHRkb2pETF0pZnpTLsqRcnl+czBsmKdzZvQP39nKnh4etcqZBoV3ywe4POHzusFbH7GTfiVf6vHLDPklJSSxatOi66a2feOIJtm7dyp133sk999wDyGkt9u3bp1VbG8tN9xAkSdoMnLuqLVaSpLoK8zsB99rHY4FFkiSVS5KUCRwF+ggh3ABrSZJ21K4KFgJ3NdbIrk5deaLrE6w6tor12esbe1m7pKOtGdOHB7DxpXB+f7IfESEurNh7grv/t52IOZv5btMxCora7ya9gkJLsmXLFsaNG4e5uTnW1taMGTNG3ybdEG3sITwGLK593BFZIOrIrW2rrH18dXujmdJ9CptzN/P2jrfp7twdRzPdFopo7ahUgv5+DvT3c+DtMZX8dSCPJYm5vL/uMB/GpBEe6MS9vTwID3LCVKO4lBTaNjf7Ja9LWtOqvFlRRkKI14Aq4Ne6pga6STdov964k4UQiUKIxNOnTwOgUWl4f/D7XKq8xNvb31bi8G8BK1MNE3p7snzqANa/OJQnB/ty4EQhT/2SRJc3Yxjz1VbeWpXCyn0nyDlXovzbKihoiSFDhvDnn39SWlpKUVERq1ev1rdJN6TJKwQhxMPIm83D620O5wIe9bq5Aydr290baG8QSZLmAnNBznZa1+5n68e/e/6bD//+kBVHVzAuYFxTzW+3+DlZ8uqoTrw0MpBtx86yK+Mse46fZ/HfOczfngXIVd/CPG3p4WlHmKct3dxtlY1pBYUmEBYWxoQJEwgNDcXLy4vBgwfr26Qb0qj010IIb2BNvSijKOBTYKgkSafr9QsBfgP6IG8qrwcCJEmqFkL8DUwDdiFvKn8pSdJfN5v76vTXNVINT8Q+QcqZFJaPWY67lfsNrlZoLFXVNRw+VcTenAvszT7PnuPnyTpbAsjpNDq7WRHmaUeYpx09PG3xtDdvVUthhfaHoUQZtQTaijJqTNjp70A44AjkA28iRxWZAGdru+2UJOmp2v6vIe8rVAH/liRpXW17L/4JO10HTGtM2GlD9RBOFp9k/KrxBNkH8ePIH1GrlF+vuuBscTn7ci6w5/h59mRfIDn3AiUV1QA4WhoT6iGLQx8fe3p52SkCoWBQKIKgA0HQN9crkLPy6Er+b9v/8VKvl3g45GE9WNb+qKqu4Uh+sSwQx8+z7/gFMs5cAiDUw5aXI4MY6K9s9isYBoog6OYcgkEyxm8MG45v4PM9nzOgwwAC7AL0bVKbx0itIriDNcEdrHmgnxcA5y9VEJ1yii/Xp/OvH3bR39eBlyKD6Ollp2drFRTk4vZtfeWqzR/1Bp/L6HoIIXhzwJtYGVvx363/pbJaSfKmD+wsjLm/jycbXgrnjTuDSS8oYvw323ls/t+knCzUt3kK7RhTU1POnj3bpqPmJEni7NmzmJqaamW8VusyqiPheALTE6bzZNcnmR42vQUtU2iIS+VVzN+exXebjnGxrIo7urnxQkQgfk6W+jZNoZ1RWVlJbm4uZWVt+yCmqakp7u7uaDSaK9rb1R5Cfd7Y9gYrj61kQdQCQp1DW8gyhRtRWFrJ95sz+GlbJmWV1dwd5s5zwwOUpHsKCi1EuxWE4opi7ll9D2qhZunopZhrlC8dQ+FMcTnfbDzGzzuzkSSJ+/t48uwwf5yttbPEVVBQaJimCEKr3UOoj6WxJe8OfJecohw+TfpU3+Yo1MPR0oTX7wxm08vh3NvLg992HWfIRwm8v+4Q5y9V6Ns8BQWFerQJQQDo7dqbh4IfYnHaYrae2KpvcxSuws3GjFnjurL+xaGM6uLG3M0ZDPkwgc/ij1CkVH1TUDAI2oTLqI7y6nImrplIYXkhf479ExsTGx1bp9BU0k4V8WlcGjEp+diZa5ga7sdD/b2VRHsKClqi3bqM6jBRmzBr0CzOl51n5k6lboIhE+RqxXcP9mLVswPp6m7LrL8OM/SjBBbtPk5VdY2+zVNQaJe0KUEA6OzQmamhU1mXtY51mTetv6OgZ7q527LwsT4sntyPDrZmvPrHAUZ9voW41Pw2HT+uoGCItDlBAHisy2N0c+rGezvfI/9Svr7NUWgEfX0d+GPqAL59IIzqGoknFyZy33c7SMo+r2/TFBTaDW1SEIxURswaNIvKmkre3P6m8kuzlSCEIKqLGzHPD+G9u7qQeaaE8d9s56mfkzh2uljf5ikotHnapCAAeFl78WLPF9l2chtL0pbo2xyFW0CjVvFAPy82vRzOCxGBbEk/zcg5m3ntzwNK+U8FBR3SpqKMrkaSJKbGT2VPwR7i7olToo5aKWeKy/lyfTq/7jqOsZGKJwb7MnmIL5YmrTY3o4KCzmn3UUZXI4RgWtg0SqtKic+O17c5Ck3E0dKEt8d2If6FoQzr5MwX69MZ+mECC7ZnUVGlRCQpKGiLNi0IAMH2wXhYeRCdFa1vUxSaibejBV9PCmPlMwMJcLHkzVUpRMzZxJr9J5V9IgUFLdDm19xCCKK8o/jx4I+cLT2Lg5mDvk1SaCbdPWz5/cl+bEw7zex1h3n2t718757Bq6M609/vxv+/ZZXVFJZW/nMrqbzy79JKLpZWUlZVjYWxEZamRliZGGFlqsHS1AhLk+u0mRihVrXtvPsKbZ+bCoIQ4ifgTqCgXk1le2Ax4A1kAfdJknS+9rkZwONANTBdkqSY2vae/FNC8y/gucaU0NQGUT5RfH/ge+Kz45nQaUJLTKmgY4QQDOvkzJBAJ/7ce4JPY9O4//udDAl0wtfR4pov+brbzVxMVqZG2JhpMDFSUVJRTXFZFcUVVTTmnWpurMbSxAgrUyMsTTVYmRhxWydnHhngjUoRC4VWQGNqKg8BioGF9QThQ+CcJEmzhRCvAnaSJL0ihAgGfgf6AB2AeCBQkqRqIcRu4DlgJ7IgfFFXb/lGNGdTuQ5Jkrhr5V3Ym9ozL2pes8ZSMEzKKqtZsD2LuZszqKiuwcZM0+DNuoE2W3P53spU0+Cv/JoaiZJKWRyKyiopKq+ShaL2vqhcbq9rq3s+/2IZh08VMSTQiU/u7Y6TlYke/mUU2is6KaEpSdJmIYT3Vc1jgfDaxwuAjcArte2LJEkqBzKFEEeBPkKILMBakqQdtYYuBO4CWuQocZ3b6JvkbygoKcDZ3LklplVoQUw1aqYM9WPKUD+tj61SictuIVebxqftliSJX3cd5901qYz6fAtzJnRncICT1u1TUNAWTd1UdpEkKQ+g9r7uG7YjkFOvX25tW8fax1e3txiRPpFISMRmxbbktArtGCEED/TzYtWzg7Az1/Dgj7uZve4wlUquJgUDRdtRRg05SqUbtDc8iBCThRCJQojE06dPa8UwXxtfAu0ClWgjhRYnyNWKVc8O4v4+nny76Rj3fruDnHMl+jZLQeEamioI+UIIN4Da+4La9lzAo14/d+Bkbbt7A+0NIknSXEmSekmS1MvJSXtL7CjvKJJPJ5NXnKe1MRUUGoOZsZr37+7K15PCOHa6mNs/38Lq5Ot+BBQU9EJTBWEV8HDt44eBlfXaJwohTIQQPkAAsLvWrVQkhOgnhBDAQ/WuaTGivKMAiMmKaempFRQAuKObG39NH0yAiyXTft/LK8v2U1JRpW+zFBSARgiCEOJ3YAcQJITIFUI8DswGIoQQ6UBE7d9IkpQCLAFSgWjgGUmSqmuHmgr8ABwFjtFCG8r18bD2INghWHEbKegVD3tzFk/pz9PhfixJymH0l1s5lHdR32YpKLTtXEYNMf/gfD5J+oS/xv2Fh7XHzS9QUNAh246e4d+L91FYWsnrd3TmgX5eyItoBYXmoeQyagSR3pEAxGQrbiMF/TPQ35F1zw1mgJ8Dr69M4alfkrhQUqFvsxTaKe1OENws3eju1J3oTMVtlFmYSXGFUmdA3zhamvDTw735vzs6s+FwAbd/voXdmef0bZZCO6TdCQLIm8tp59PIKMzQtyl6I+1cGuNXjefjxI/1bYoC8uG3Jwb7snzqADRGKibO3cHn8elU1xi2S1ehbdEuBWGk90gEgpjM9uk2Kq8u59Utr1JZU0lcdhyV1ZX6Nkmhlm7utqyZNogx3TswJ/4Ik77fSV5hqb7NUmgntEtBcDZ3JswljOis6HaZNvmLPV9w9MJRJgZN5GLFRXbk7dC3SQr1sDLVMGdCKB/f250DJwq5/fMtrNmvnFlQ0D3tUhBAdhtlFGaQfiFd36a0KDvzdrIwdSETgibwn97/wcrYSjmXYYAIIbinpztrpg3Cw96cZ3/by9O/JnGmuFzfpim0YdqtIIzwGoFKqNrV5nJheSGvbX0Nb2tvXuz1Ihq1huGew9lwfAMV1UpkiyHi62TJH1MH8J+oIOJTC4j4dBOrk5WCQAq6od0KgqOZI31c+xCTFdMuPlySJPHuznc5V3qO2UNmY2ZkBshhuMWVxWw7sU3PFipcDyO1iqfD/Vk7fRCeDhZM+30vU3/Zw+kiZbWgoF3arSCA7DY6XnScQ+cO6dsUnbMmYw0xWTE8Hfo0IQ4hl9v7uvXFxsRGOZfRCghwsWL5U/15dVQnNqQVEDFnEyv3nWgXP2gUWoY2X0LzRozwGsF7O98jOiuaYIdgfZujM04Wn2TWrln0cO7BY10eu+I5jUrDCM8RrMtcR1lVGaZGjc/3r9AC1FRD8u9QUQJCYCRUPGWh4u4RFSxOPMnWpXGc22bL+F6eWJuZgFCBELX3KqDeYwd/cArU9ytSMGDatSDYmNjQr0M/YjJjeD7s+TaZMqC6ppr/bv0vEhKzBs1CrVJf0yfSO5Ll6cvZdmIbw72G68FK7VBaVcoLG1/gya5PEuYSpm9ztEPaX7DymWuanYFpABrgNI3LDGbuAC+lQwPvAQUFaOeCALLb6P+2/R/7z+ynu1N3fZujdeanzCcpP4n3Br6Hu5V7g316u/bG3tSe6KzoVi0Im3M3s/XEVmqkGr6L+E7f5miHlD/lL/KndwICpJprbllnivlg3SEO5V1goK8DL4zww8FCc2W/jI0Q9wbk7YOOPfX9qhQMlHa9hwBwm+dtaFSaNhltlHo2la/2fUWEVwRj/MZct5+RyogRniPYlLuJksrWW7ilLnx2+8ntZF/M1rM1WqCyFNKiofNosHQGSyewcgFrN7DpCLYeYOeFd0AIXz07nn+Nuo1l2abctvAUf+RaITkHg2tXcOsO3e+Xx8zYpN/XpGDQtPsVgpWxFQM7DiQ2O5aXe7+MSrQNjSyrKmPGlhnYm9jzRr83buoOi/KJYsmRJWw5seVyAsDWxKXKS2zO3UyEVwQJxxNYkraEl3u/rG+zmsfR9VB5CYLH3rSrWiV4cogvt3V25j/L9vPCkmTW7s9j1t1dcbE2lQXFOUReKQx+4abjSZLEmeIKjhYUc/R0MccKijl2upiM05cor6pGJQRqlUAlBCoVqIVApRKob9Be/3lzYzWDAhyJCHbBzcZMC/9YCtqg3QsCyG6jjTkb2Vuwl54ubWM5PSdpDhmFGXwX8R22prY37R/mHIajmSMxWTGtUhASchIory7ngc4PIBCsOLqCaT2mte5N8tQVYGYP3kMafYmfkyVLpvRn3rZMPopJI+LTTbw5OoS7wzoifMPh7x/klYdG/hKurpHIOVfCsdPF8pd/7Rf/0YJiLpb9U7jH3FiNn5Mlvb3tMDcxoqZGokaSqK6h9l6iWpKQ6h7Xa6+RpH/61EhUVNeQc76E2NR83liZQteONowMdiEixIUgF6s2uZfXWlAEAQj3CMdEbUJ0ZnSbEIStJ7by2+HfeKDzAwzoMKBR16hVaiK8Ivgj/Q9KKksw15jr2ErtEp0ZjYu5C6HOoVTVVBGbHUt0VjR3+d+lb9OaRmWZ7C7qMg7Ut/YxVdcmyhve2YWXlybz4tJk1h7IY4Z/TwKqv2bpiuVsrAiWf/GfuURFVc3lax0tTfB3tmB09w74O1vi72yJn5MlbjamWv+iPlpQTGzqKeJS8/kk7gifxB3Bw96MkcGuRAS70MvLDiN121ixtxYUQQAsNBYMcR9CXHYcr/Z5tcFInNbC+bLzvL7tdfxt/Xku7LlbujbSO5LfD//OxpyN3O57u44s1D6F5YVsO7mNSZ0moRIqerv2xsfGhyVpS1qvIBxbDxVFENx0+30cLVg8pT8LtmfxYcxhxh6GZBM1Z5JjSLHpiJ+TJUMDnfBzssTP2RJ/J0tszDVafBE3RhYcf54O96fgYhnxhwqISz3Fzzuy+XFrJnbmGm7r5EJEsAtDAh0xN1a+rnSN8i9cS5R3FHHZcSTmJ9LXra++zWkSkiTxzo53KCwv5NsR396yu6SHcw+czZyJyYppVYKw4fgGqmqqGOUzCpDzAE0ImsDs3bNJOZtyxUG8VkPKCjCzA5/Gu4saQq0SPDbIh4hgFw6eKKRyWxhTRA5TpwzTkqHawdnalEl9PZnU15Pi8io2HzlNXGo+camnWL4nFxMjFYNr9xyGd3bB0dJE3ya3SZq1HhNCPC+ESBFCHBRC/C6EMBVC2Ash4oQQ6bX3dvX6zxBCHBVCpAkhDMpRPdh9MGZGZq263vKKoyuIPx7PtB7TCLIPuuXrVULFSO+RbD2xtVUVzonOisbd0v2KL/4xfmMwMzJjSdoSPVrWRKrKIW0ddLoD1Nr5xe5hb86orm6YdxqBKm8flBhuAR5LEwDXSZAAACAASURBVCNu7+rGnAmhJL0ewW9P9OX+Pp4cyivileUH6D0znnu+2c53m45x8EQhZZXVNx9UoVE0WRCEEB2B6UAvSZK6AGpgIvAqsF6SpABgfe3fCCGCa58PAaKA/wkhDMY3Y2ZkRrhHuFwfoKb11QfIKcph9u7Z9HbtzUPBDzV5nEjvSCpqKkjISdCidbrjXNk5duXtIson6goft5WxFbf73M5fGX9RWF6oRwubwLENte6icdof22coIEHWVu2PrQM0ahUD/B15a0wIW18Zxtrpg5h+WwAlFdW8v+4wd365leA3ornt441M+TmRT2PTWJ18kiP5RVfsjSg0jua6jIwAMyFEJWAOnARmAOG1zy8ANgKvAGOBRZIklQOZQoijQB/AYJLxR3lHsS5zHbvydjGo4yB9m9Noqmqq+O+W/6IWamYOnNmsPZBuTt1wtXAlJiuG0X6jtWilbojPjqdaqibKO+qa5yYETWB5+nJWH1vNA8EP6MG6JpKyAkxtwXeo9sd27wXGlnL4afD1z6YYIkIIQjrYENLBhucjAsk9X8K+nAscOVXEkfxijuQXEZeaT12ROSOVwNfJggAXK4JcrAh0sSTQxQovBwvUKiWSqSGaLAiSJJ0QQnwMHAdKgVhJkmKFEC6SJOXV9skTQjjXXtIR2FlviNzatmsQQkwGJgN4eno21cRbZlDHQVhqLInOjG5VgvDjgR/Zd3ofswfPxs3SrVljqYSKSK9Ifj38K4XlhdiY2GjJSt0QnRWNj40PgXbX5ujp7NCZbk7dWJy2mH91/lfrCGesKpfTVXQeozV30RWoNeA1UBaEVo67nTnudubQ7Z+2sspqjp0uJj2/mLT8ItLziziQW8ja/XmX+xgbqfB3spQFwlUWi64dbXC2bsUhylqiyYJQuzcwFvABLgBLhRA3+hnW0KexwTSNkiTNBeYC9OrVq8VSORqrjbnN87bL9QGM1cYtNXWTOXjmIN8kf8Mon1Hc4XuHVsaM8oliQeoCEnISDDpKp6CkgMRTiTzV/anrftlPCJrAa1tfY/ep3a0jWCBjI5RfbNRhtCbjGw7pMXAhRz7t3IYw1agvryLqU1JRxdGCYtJOFZFee7878xwr9v1Tia6jrRmhnrb08LClh6cdIR2sMdUYjFe7RWiOy2gEkClJ0mkAIcQfwAAgXwjhVrs6cAMKavvnAvXffe7ILiaDItI7klXHVrH95HbCPcL1bc4NKaksYcaWGTiZO/Fa39e0Nm6IQwgdLTsafBx/XHYcElKD7qI6Ir0j+fDvD1mctrh1CELKCjC1kb+0dUWdKypzE/RoRa60ZmBubEQ3d1u6uV95SPNiWSVHThWxL+cC+3IusPf4hcurCY1aEOxmTQ9PO3p42tLDww4Pe7PWsdJsIs0RhONAPyGEObLLaDiQCFwCHgZm196vrO2/CvhNCPEp0AEIAHY3Y36d0N+tPzYmNkRnRRu8IHyS+AnZF7P5YeQPWnXtCCGI9I5kYcpCLpRdaNRJZ32wLnMdgXaB+Nr6XrePidqEu/3vZmHqQgpKCnA2d75uX71TVQFpayHoDjDS4erUORgsnOTVSDsRhOthbaqhl7c9vbztL7cVXCxj72WBOM+SxBzmb88CwN7CuHYFIa8iurnbYGXacmc3dE1z9hB2CSGWAXuAKmAvspvHElgihHgcWTTure2fIoRYAqTW9n9GkiSDixfTqFtHfYBNOZtYcmQJj4Q8Qh+3PlofP9I7kp8O/sT64+sZHzhe6+M3l7ziPJJPJzO9x/Sb9r038F7mpcxj+ZHlTA2d2gLWNZGMjVBWCCE6XpUJIa9AMjaBJMl/K1zG2dqUyBBXIkNcAaiqruFIfvFlgdibc4H1h2XHhxAQ4GxJDw87unvYEtLBmiBXq1brampWlJEkSW8Cb17VXI68Wmio/0xgZnPmbAnq6gNsPbGVEV4j9G3ONZwtPcsb298g0C6QaT2m6WSOzvad8bTyJDor2iAFoS6z6Y3cRXV4WHswsONAlh1ZxhPdnkCjMtBfdKkrwETH7qI6fMPhwFIoOAQubbc4lDYwUqsI7mBNcAdrJvWVg1wKSytJrreKiEk9xeLEHABUQs4pFVJ7TbCbDcEdrLG3MPw9SeWkcgPU1QdYl7nO4AThUuUlnkt4juKKYn4Y+YPONr7r3EY/HvyRc2XnsDe1v/lFLci6rHWEOITgYd24TdEJgROYnjCdTTmbDO7/FJDdRYfXQNAoMGqBU7g+tfsIGRsVQWgCNmYahgQ6MSTQCZCzBOScKyU1r5DUkxdJzbt4zaa1q7VpPZGQ7z3szFEZUAisIggNYKQyIsIrgpVHVxpUordLlZd4Ku4pDp45yEdDPyLALkCn80V6R/L9ge+Jz47nvqD7dDrXrXD84nFSz6byUq+XGn3NEPchuFm4sShtkWEKQubmlnEX1WHrAfZ+siD0f7pl5mzDCCHwdDDH08GcqC7/hH6fv1RBat7FyyKRevIiG4+cprr2sISliRGd3awIdrMmpIO8kvB3ttSby0kRhOsQ6R3J4rTFbMrddDlHjj65VHmJqfFTOXDmAB8N/YgIrwidzxloF4iPjQ8xWTEGJQh16UVuJU23WqXmnsB7+HLvl2QWZuJj46Mr85pG6p9gYg1+t7XcnL7hsH8xVFfq5syDAnYWxgz0d2Sgv+PltrLKao7kF10hEsuSclmwQy7qpBLg7WBBgIslQS5W8sE6Vyt8HC3Q6Dj7qyII1yHMOQwnMyeiM6NlQdj6GZxIAntfsPepvfcFqw6g0u1/0qXKSzwd/zT7T+/nwyEftogYwD9uo7n753Km9AyOZo43v6gFiM6KpodzD1wtXG/pursD7uab5G9YkraEV/q8oiPrmkB1JRxe23Luojp8h0Lij/L72rNfy83bzjHVqK8Jga2pkcg+V0LqyYuknbqot5PXiiBcB7VKzUjvkSxNW0rxuQwsN7wr/4I7Eg3VFfU6mlwpEPY+YFf7t43HLeeyv5qSyhKejn+a5NPJfDDkA0Z6j2zmK7s1Ir0i+Tb5W2KzYpnUeVKLzt0Qxy4cI/18Oq/2efWWr3U0cyTCM4KVx1YyPWw6ZkYGUqkrcxOUnm9Wqusm4T0YELLbSBEEvaJSCXwcLfBxtOCObv+4nBo6eb0/98INT14HOssriqagCMINiPKO4tdDv5Kw82NG11TB47HyF/3FE3Auo94tU74/lgBVpf8MoDICW696YuELTkHyUr0RoX4llSVMjZ9K8ulkZg+ZrZdKZv52/vjb+hOTFWMQghCdFY1AMNKracJ4X9B9rMtaR3RmNOMCdJA8rimkrgRjq5Z1FwGY20OHUFkQwm9dYBV0z/VOXl8ql09eH8kvqr0Vs+uqTeymoAjCDeju1B03CzeiT2xhtPdgcKzdxLX1lG++4VdeUFMDxaeuFYpzGXB8B9SllL7tdRhy4w3RksoSnl7/NPtO7+ODwR80KrxSV0R6R/K/ff8j/1I+LhYuerNDkiSiM6Pp7dobJ3OnJo3R06Un/rb+LE5bbBiCUF0Jh9ZAUBRo9HDmxTcctn8J5cVgYtny8ys0CQsTI7p72NLd49qT1+m1AjHpg1sfV6lPdwOEEETadGK7RqIwdMLNL1CpwLoDeA+CsIdgxJtw3wJ4agvMyIWXjkLn0bDpAzn++zrUicHegr3MHjybKB/9iQHIgiAhEZcdp1c70s6nkXUxq1krJSEE9wXdR8rZFA6eOahF65pI1hYoPdfy7qI6fIZCTRVkb9fP/HqmvLqct7a/xfyD88m/lK9vc5qNtamGnl723N+naUlBFUG4CVGnc6gSgvWmzdzsEwIsneCOOWBiBSuehuqqa7qVVJbwzPpn2Fuwl/cHvW8QEU4+Nj4E2QXpvXhQdGY0aqFu9qb6aN/RmBmZsThtsZYsawYpK+R01P4NnuXUPZ795H2wNpD9tCmsPLqS5enL+STpEyKWRfBE7BP8mf5nqyoQpU0UQbgRF08SnL4Jd7UF0cfjtTOmpRPc/hGc3AM7vrziqZLKEp7d8Cx7CvYwa9AsgypjGeUTRfLpZPKK827eWQdIkkR0VjT93PphZ2p38wtugKWxJXf63sm6zHX6LZ5TXSUfRguMBI2eNrg1ZrIotENBqK6pZn7KfLo6dmX1XauZ0n0KJ4pO8Mb2NwhfEs5Lm15iY85GKqtbX8GspqIIwo3Y8zNCqibKZxS7T+3mXJmWyg6G3C27jhLeh9NpAJRWlTJtwzSS8pOYOWim1lJZa4tIL9lNE5sdq5f5D545yIniE1pzn00ImkB5dTkrj668eWddkb0VSs7qz11Uh+9QKEiB4oKb921DxB2PI6coh8e6PIa3jTfPhD7DX3f/xS+3/8I4/3HsytvFtA3TGLZ0GO/tfI99BfuQpBbLxq8XFEG4HtVVsGcB+A4jqvNEqqVq1mWu087YQsAdn4KxOax8htKKYqatn0ZifiLvDXyPO33v1M48WsTD2oNgh2CiM/XjNlqXtQ6NSsNtntqJxAmyDyLUKZQlR5ZQI+mp1GLKCtBYQEDLnCu5Lr7h8n3mZn1a0aJIksS8g/PwtvZmmMewy+1CCLo7dee1fq+x4b4NfD38awa4DWDl0ZU8uO5BRv0xii/3fklGYYYerdcdiiBcj6Nxcnhpr8cItAskyC6I2btnM37VeL5N/paMC818Q1g6w6iPKD2RyLQV97D71G7eG/ieQZetjPSO5ODZg+QW5bbovDVSDTFZMQzsOBBrY2utjTuh0wSyL2azK2+X1sZsNNVVcGi1ft1FdbiFyjUYMlpHHW1tsOvULlLPpvJIyCPXLTmrUWkY4j6ED4d+yMYJG5k5aCaeVp78cOAHxq4Yy4Q1E/g59WfOlJ5pYet1hyII1yNxHli6QNAohBDMHTmXV3q/gqXGkv/t+x9jV47lrhV38fW+r0k/n96kpWRZ5zuZ7tOZ3SW5zOw+3aDFAP5JFVGXabSl2Fuwl4KSAq2H3o70GomdiZ1+Npezt0HJmZbLXXQjVGrwGfJPOux2wE8HfsLRzJE7/Rq3GrfQWDDGbwxzR84l/p54Xu71MpIk8eHfHzJ86XCeinuKFUdXtPpIJeUcQkNcOA7psTD4xcs5XuxN7Xkg+AEeCH6AgpIC4rPjicuO47vk7/g2+Vu8rb2J8IpgpPdIguyCblpVqayqjGkJ09klXeK9wjJGJy2Fbo/KH04DpaNlR7o6diUmK4bHuz7eYvNGZ0Zjqja9YmmvDYzVxowLGMeClAWcunTqllNhNIvUlaAxB389u4vq8Bkqr1jOZYCDn76t0SmpZ1PZkbeDf4f9GxP1rUcPOpk78VDIQzwU8hAZFzJYk7GGvzL/4vVtrwPgbe1NH9c+9HHrczlzcmtBEYSG2LNQvu/5cINPO5s7M6nzJCZ1nsSZ0jNsOL6B2OxYfjz4I98f+B4PKw9ZHLxGEuwQfI04lFWVMX3DdHbl7eLdge8y5lIp/DkFdn1n8JknI70j+TjxY45fPI6ndW2sc2WZnINHB4VWqmqqiM2OZbD7YJ1knb038F7mHZzH8vTlPBP6jNbHb5CaavnLN2CkvI9kCPjWim3GxjYvCPMOzsNSY6mVhI2+tr5MD5vOtB7TSDufxu683ew6tYs1GWtYcmQJICeJ7OvWl76ufenp0hNLY8M9AKgIwtVUV8Ken+WNPtubH+5wNHPkvqD7uC/oPs6VnSPheAKx2bEsSFnATwd/ooNFByK8IojwjqCrY1cqqit4LuE5dubt5J2B7zDWf6y8TD/4B6x/R/YpG/AHsk4QYrJieLLbk3K6jp/HgVCBmS2Y2d3azdTmhquixPxEzpWd09lJbXcrdwZ1HMTyI8uZ3G1yyxTPyd4OlwoMw11Uh4MfWLvLgtC75VZ/LU1OUQ6x2bE8HPIwVsZNy/fTEEIIOtl3opN9Jx4KeYjKmkpSz6ZeFoglaUv4OfVn1EJNiEMIfdz60Me1D6HOoYaTU4tmCoIQwhb4AegCSMBjQBqwGPAGsoD7JEk6X9t/BvA4UA1MlySpZZ3RjSFtnZx+otdnt3ypvak94wPHMz5wPBfKLpCQk0Bcdhy/Hv6VBakLcDF3wd7UnsPnDvP2gLf/KWAvBIz+DL7uByufhUfW6jyDalNxtXAl1CmU6KxoWRASf5Jz4vR8VE7QVncrLoAzR+THZTeJ9Te1kcUh5G75dHc9ojOjMTcyZ7D7YJ29pomdJvLM+mdIOJ7QMskDU1eAkZm8QjAU6spqpq2VVzC6dF1WV8KJPeDRp8XLdy5IWYBaqHmgs25rSWtUGro7dae7U3ee7PYk5dXlJBcks+vULnbn7Wb+wfn8cOAHNCoNoc6h9HHtQ1+3vnRx6IJGj6nIm7tC+ByIliTpHiGEMWAO/BdYL0nSbCHEq8CrwCtCiGBgIhACdADihRCBBldXOWkeWHdstm/X1tSWcQHjGBcwjosVF9mUs4nY7FiSTiXx9oC3r82jY90BombBymfg7++h75Rmza9LonyimL17Nhn5+/A9Eg29Hofhr1//guoqKL94pWBcfcveDju/gaH/uRx1U1lTSfzxeMI9wnX6K2pgh4F0sOjA4rTFuheEmmpIXSWvQI0tdDvXreI7FPb9Aqf2Q4ceuptnw7uw7XMY8xWEPai7ea7ibOlZVhxdwWi/0TibO7fYvAAmahN5VeDWB3rIKe335O9h96nd7Mrbxf/2/Y+v932NmZEZPV160s+tH/3c+hFgF4BKtNyPwyYLghDCGhgCPAIgSVIFUCGEGAuE13ZbAGwEXgHGAoskSSoHMoUQR4E+wI6m2qB1zmXCsQ0QPqPZaavrY21szWi/0TePIgr9F6T8CfFvyb8e7Q2siEstEV4RfLD7A2KS/sfU6groPvHGF6iN5FWE+Q0219Lj4dfxkLUNAuSKZjtP7qSwvFDn6TvUKjX3Bt3L53s+J+NCBr62vrqb7PhOw3MX1VG/rKauBOF8Nuz8Vs4EHD1Djm6y89LNXFfx2+HfqKiu4JGQR1pkvhthobFgsPvgyyvfC2UXSMxPZGfeTnbl7eLjEx8Dstehr1tf+rv1p59bP9ws3W40bLNpjvT4AqeBeUKIvUKIH4QQFoCLJEl5ALX3dVLcEcipd31ubds1CCEmCyEShRCJp0+fboaJt0jSfBBqOTGdPhACRn8uf1hWTZOzpxogzubOhLmEEVPwNzh1ArfuzR/Ue6DsRkn/5yR0dFY0VhorBnQY0Pzxb8I4/3FoVJrLG4E6I3UFGJlCQMunMr8pVi7gHKzbNBYb3pXf5w+vkf9e+UyLvM9LKktYdHgRt3neZnjV8pA9CiO8RvB//f6P1eNWE3dPHO8OfJf+HfqzO283b2x/g5HLR3Lnn3fy3s73iM+O10naleYIghEQBnwjSVIP4BKye+h6NOQsbDDoWZKkuZIk9ZIkqZeTU9PSHN8yVRWw71cIjJLdN/rCxh1GvidnwUz6SX923IRIp54cE1Uc7RypHT+wxkz+tVgrCOXV5Ww4voHhXsMxVhs3f/yb4GDmQIRXBKuOrqKkskQ3k9TU/OMuMtRU0z5D5VVMZZn2x85NggNLof8z4NVfdpFmbYHdc7U/11UsT1/OxYqLPNrlUZ3PpQ1cLVy5y/8uZg+eTcJ9Cfwx5g/+0/s/eFl7serYKp7f+DxDFg/h/jX38/mez9mVt4vy6vJmz9scQcgFciVJqjvmuQxZIPKFEG4AtfcF9fp71LveHWheNQdtcngNXDoNvQzgDRP2kBwGGPuGvMQ2QCLOn0YlSUSba9G3HxAB5zPh7DG2nthKcWVxi9aBmNhpIkWVRdpLUXI1OTvlgAV95y66Eb7hUFUGOVo+vS1JEPt/YOEEg56X23o8KK+U4t+EM+nana8elTWVLExdSE+XnnR30sJqtoURQhBgF8CDwQ/y9fCv2TZxGwuiFjCl2xSMVEbMOziPJ2KfYODvA5kcO5mfDv5E6tnUJs3VZEGQJOkUkCOECKptGg6kAquAugD+h4G67GGrgIlCCBMhhA8QAOxu6vxaJ/EnOcy0patWNYQQMOZL+X7VNMM7PSpJOKaspLcwIyZvu/YSfvnLewekxxKTGYOdiZ28CddChDqFEmAXwOK0xbpJYpayQk41HWiA7qI6vAfKblNtu40Or4Hj2+X9OZPacE8hYMwX8urwzykNpoPXBusy13Hq0ike6/KYTsZvaTRqDWEuYTwd+jQ/3/4zWydu5cvbvuSewHs4XXqaOUlzmLCmEfVbGqC529fTgF+FEPuBUGAWMBuIEEKkAxG1fyNJUgqwBFk0ooFnDCbC6MxReeka9rDhnBS29YCR78r1dpPm69uaK8nZBeezGOkeTtbFLI6cP6Kdce19wDGQkiPRbMzdyAivES1zLqAWIQQTgyZy6NwhDpw5oN3Ba2rgUJ27SHvx71rHxArce2tXEKoqIO4NcAySP2P1sXKFOz6BE0mw7dZDvW9GXRI7f1t/BnfUXeiyPrE0tiTcI5xX+7zKn2P/ZP2965k1aFaTxmpWKI0kSfuAXg081WC1D0mSZgIzmzOnTkiaJ2/k9mi5ELhG0fNROeoo9nX517Otx82vaQmSF4HGnBG9pjPrRBzRWdEE2Qfd/LrGEDCSzQcXUupoq5fiQHf43sEniZ/wS+ovTOsxjUqpksrqSqpqqqisqbx8q/93VU0VldUNt4d7hNPJvhPk7oaiPMN2F9XhOxQ2fySHA5s1r/YEIH++zmXApCUNR+91GS+f3N44W149uXZt/py1bDmxhaMXjjJr0KybppNpKzibOzc5L5pyUrmyTN5M7nSHHGVhSNS5jv43AFZPhwf+aPGDPNdQWQYpf0Dn0djbeNDHtQ8xWTFM7zFdOx+4gAhiji3CUWNFmHNY88e7RSw0Foz2G83itMWsy2r+XsLOvJ3Mj5r/j7soSL/lUBuFb7hc5jVrq1y3ozmUXpC/6H2G3Pgg3h2fymdR/pgCkxPkVCha4McDP+Jm4ab3MrStBUUQDq2Sfwn1NIDN5Iaw84aIt+Gvl2Dvz/oLia3jSLR88rib7KO83fd2Xt/2OpPWTuLhLg8zwnMERqqmv62K3bqy2dyMe9S2101LrGum9ZhGF8cuqIUajUqDkcoIjUrzz2O15or2a55XadCoNXy19yt+OfQLJeXFmKeulFd5huwuqqNjL7lOQ8bG5gvClk/kz9fImTf+MWNuD6O/gN8nyAJy1Yn1prCvYB97CvbwSu9XWtT12JpRBCHxJ7D3/edQjiHS63E5O2bMa+A3HGwaPL7RMuxfDJaul4uqjPEbQ0V1BQtTF/LyppfpYNGBB4If4O6Au7HQ3PpJ3IST26kQglEFOfJmuh5WRDYmNv+kFWkG/Tv0Z37KfJJSFzG46CQEv9V841oCI2PwGiCnw24O57Ng17fQ/X5w63bz/kFR0OMBeS8haJSc2qIZzDs4DxsTG+4OuLtZ47QnDDNhTktRcAiO74Cejxhs7iBAtm3MF1BTBauf01/U0aUz8jmBbvdd3nxXCRX3Bd3HyrEr+WzYZ7hauPLh3x8SsTSCT5M+5dSlU7c0RXRWNK4aK7qdPX65vGhrJcw5DGOVMTuOrga1cetwF9XhGw5n06GwGcWQ1r8jRyzd9n+NvybyfTnJ3p9PQcWlJk+dUZhBQk4CE4Mm6iRLblvFgL8FW4Ck+fIHNfRf+rbk5tj7wvA35Upuyb/rx4aDf8ii1ECqCrVKzXDP4SwYtYBfb/+V/h36syBlAaOWj2LGlhkcPnf4psMXlhey/eR2orxGym/MdP3Ub9YWpkam9HAOZefFY/LKztRG3yY1Ht9w+b6pq4TcRDi4HAY8e2srWlNruOtrOHdMTuHSROYfnI+J2oRJnSc1eYz2SPsVhIoS2Pc7dB4DFo76tqZx9JkMngMg+lW4mNfy8yf/LkeAuITcsFs3p258Ev4Ja8etZWKniaw/vp57V9/LEzFPsCV3y3Vj/NcfX09VTRVRQffKKRSOxuniVbQo/cw9SDcSnAkcoW9Tbg3nYPkQWVPCTyVJdm9aOMPA5279ep8h0HeqfIK5CfPnX8pndcZq7vK/q1UVpzEE2q8gpPwJ5YWGcTK5sahUMPYrqCqHtS+07Nynj8DJPdDtJons6uFu5c4rfV4h7p44/h32bzILM3l6/dOMWzmOP9L/uOaofXRmNB5WHgQ7BMvx+tk7oOyitl9Ji9K/UM7FtcuyFa0OQH6v+QyRz8Hcqovy0Gr5VPaw/zZ9E334G+DgDyueuXn69Kv45dAv1Eg1PBzScIErhevTfgUhaR44BoLXQH1bcms4+MkftLS/4Oj6lpt3/yK5CE7Xe2/5UhsTGx7v+jjR46OZNWgWRioj3tz+JpHLIvku+TsulF3gbOlZdp/aTZR3lBy+GjASairlL6TWiiTRKX0T1qjYcSZZ39bcOr7hUJwPp/9x9526dIrK6srrX1NVIaeicOrUvHM9xuYw7jsoOilnRW0kFysusvTIUiK9InG3cm/6/O2U9ikIpw5A7t9yqKm+4/qbQt+nwNZLPv1Z0wKHvWtqYP8S2Q/ejLMaGrWG0X6jWTp6Kd+P/J5ODp34at9XRCyL4PmNz1MtVf8TL+7RF0ys9bOPcPaY/CWU8D7smgsHlsmV4fL2Q+EJqCxt3Dgn9qAuzKGvTQA783bqJh2GLvENl+9r3Tbl1eWMWzmOd3a+c/1rEn+UD6GNfK/5KeTde8GgF+RzQof/atQlS9KWcKnyUqtJYmdotM+w08R58iGhm+XxN1SMTGDEW7DsUdmv30O31Z/I3gaFOfKcWkAIcbkAyNHzR1mYupA1GWsItAskwDZA7qTWgN8wuU5CS4efbnhPdik2nIxXRmMO5g61dR4c5JuZ/ZVtR6JBpaGf/2jikj4m62KWQaZevi62nmDnIwtCv6kkFyRTXFnMyqMr+Vfnf8knsOtTel4+0OYb/k9equYy9BU4EiNH13n0BQuH63Ytry7nl9RfGNBhiLtbxAAAIABJREFUAJ0dOmtn/nZG+xOE8mL5127IuBsXbDF0QsbBjq/lL6+QcbqtvpW8CIytIOh2rQ/tb+fPOwPf4fmecgbMK047+0fI5y/yU8C1i9bnbpDi07IPvO9T8q/c0vNQchZKz8n3l2/nrnx8LlO+vzpHfeAo+nveBkkfszNvZ+sSBJC/3A8sg+pKdp/ajUqosNRY8kniJ8yNmHvl/9fmj+WTySPf056AGxnDuG9hbjisfR7uXXDdsVcdW8XZsrNtJomdPmh/gnBwGVQUQa9W/qYRAiJnwk+RsjAM/Y9u5qkokYu6BN8l+3V1hJ1pAzlz6mU/bTFB2PervHfR61HZ5WHpJN8aS1XFlSLiHIy7mR0dLTuy4+QO7u90v+5s1wW+4fJ+24k9/H3qb4Ltg7nD9w4++PsDtp3cxqCOg+R+5zLlqKDQf2k1FxEg/98P+y+sf1sOZe16zzVdqmuqWZCygGCHYPq4tlyG3LZG+9tDSJwnh9Q18xSkQeDZT04tsPUzKMrXzRxpf0FFsX7ca9Zu4NoN0lso/LSmBvYskEN7nZqYrM/IWN5ncQkG70Fgbn/ZRfb3qb+pqtFNimed4TMEEJQejWf/mf30duvNhKAJeFh58EniJ1TX7WGtf1tOEHnba7qxY8B0OQvr2hcbDLnekLOB7IvZPNblsXaTxE4XtC9BOLEH8va13s3khhjxNlSXw8b3dTN+8u9g46G/aKyAkXK67dILup8ra7O8IaqDUOR+HfpRXFlMytkUrY+tU8ztwa0be7PlMyJ9XPvw/+2deXhU1fnHP28WAgSEhCWETbYQNiFAgFhEUQw/tRZFW+tWrcsjKhZw609rf/5q1eexVdSfT7VqXYorqFjR2oqIiFaFSYKsIiayh0wIe9hJ8v7+ODcyQLa5s+ROOJ/nmWfunbnnve/MnLnvPec953sS4xOZNmwaRbuKmPPDHNjkMzmXn/wmcqsNxieYUUcVh+D9244ZCquqvLTiJbq37s653WNsvofHOLkCQsHLJhk4xN3iEZ6kXW8YcaO5s91a/2zgoCgvhR8+daQqGqmqZOSCVsLaBZE/V/7LRu65/4Swmx7VaRQAi7YsCrvtiNNrLHnl60iQ+B8VaHNPzWVIhyH85Zu/sP/j30GrNHMXH0na9YbcP0LRJ8esEZLnz2Pl9pVcO/DaRhNEbCqcPAHh4G5YMRsGXRJbEgIN4czfmqTvvPvDa3fF26BVQU1GCztdsqF528h3G+3dalb1GnIlJDYPu/mU5in0T+3P1yVfh912xOk1Fl9SIgNbdftRF0hEuCv7LsoOlDGjfA2cfV901okecaPpxpp7n8lbAC+teonU5qlM6B3+QH6ycfIEhOVvwZF9sZ9MronkdnDmnVA4N3SFykCWzYTOw6BD3/DZDJb4BOgzzgSEqqrInWfp60anafivI3aKnM45LCtbxv4j+yN2jkiwL30wq5KaMbLqWAnprNQB5B6Gl9u2ZVu/KAn3xcXBRc8YccU5k1mzbTVfFn/J1f2vpnlC+AP5yUbIAUFE4kXkGxH5p7OfKiLzRKTQeU4JOPZeESkSkTUiEr2FZVVNE7PTYHOBa4qMnARtupuFzMNx4fSvhNIVRrq4sckYD/u2gj9Cs32rqkz9OPWMiAa/nPQcKqoqKCgtiNg5IsGSHaupFGHE9i3HvpH3AtNKt3AkLo6nlz8bPYfadoPzHoENX/LS57+jZUJLLsu8LHrnb8KEo4UwFVgdsH8PMF9VM4D5zj4iMgC4HBgInAc8IyLR6fDbnA+lK02ysKkkk48nsbnRf/EvhxVvhW5v+UwzamTQpaHbCpXe4wAxk9QiwbrPjHZ/BFsHcFQOe1FJbOUR8vx5JBBHlv87M08DzJyLhX+i+6ln8st+V/Bu4bsU7SyKnlNZV1Lc+yzm7inkF30m0iapiXUDNxIhBQQR6Qr8FHgh4OWLgBnO9gzg4oDXZ6rqIVVdBxQB0Rn7mf8SNGvlSocnphh0KXQeanToGyqvUBNVlbD8bXNnXsfM0KjRqoP5XJGSsch/2cwyHhDZPmgjhz005vIIPr+PwW370EL1qLbUF9NNXi73QSYNnkRyQjJPLHkiek6J8Le0LghwNTGwCl2MEGoL4Ungt0BgH0WaqpYAOM8dnde7AJsCjtvsvHYCInKTiOSLSH5ZWVloHh7YadYAPu0XsbF8YSjExZlZonuKYdEz7u2s/Qz2+r0l7ZEx3uhP7dseXrvlpWauRdaVYVvHty5yOudQuLOQbQe2Rfxc4aD8cDmrd6xmZPexkNTG1I0da2Hxc0YypdMgUpqncOPgG/l88+csLlkcFb/e+f4dZpf8h8srkui05I3GWzSqieE6IIjIhcBWVW1oh2hNfTU1/oqq+ryqZqtqdocOQcwSrYnPH4OKg7Elcx0KPc6AzJ/CF08cbd4Hy/JZZiRWXw+t8JUxHlAzDDacLH3NSSZHp36cnn46QNQunKFSUFpAlVYxMj0Heo4xgxY++YPRmjr76CS0q/pfRXpyOtPzp1OlEUz+A18Wf8lDix5idJfR3Dn0N6Y7eMOXET3nyUIoLYTRwAQRWQ/MBM4RkdeAUhFJB3CetzrHbwa6BZTvChyXpQozS9+Er/9i/uzpQyJ6Kk+R+wAc2Q8LHwm+7KFyo+Uz8JKo3DE3mM5DoWX78C6aU1UFBTOgxxho3yd8duugX2o/Tml2SszkEXx+H83imjG4w2AjY7F7o9GXGj3VzCR3SIpPYuqwqazesZoP134YMX/W7FjDnQvvpE/bPkw/azoJg39puvsWRzGp3YRxHRBU9V5V7aqqPTDJ4k9V9WrgfaB6ZYprgTnO9vvA5SKSJCI9gQzA59rz+ti4GD6YYv7sFzwasdN4kvYZpkWU/zJsKwyu7OoPTDDxwuiiQOLizPDTok/CJ/m99lPYtSHiyeRA4uPiGZU+iq+3fB0Tcth5/jyyOmaRFJ90VA67VSczK/k4zu95PgPaDeCpb57iYMXBsPuydf9WJs+fTHJiMn8Z9xeSE5MhsYX5/b77EHZtDPs5TzYiMQ/hESBXRAqBXGcfVV0FvAV8C3wETFbVyIj579oIs66CU7rAZa+Y5u3Jxln3mFnZ8/43uHLLZhrJYy9qPWWMN6JxW74Jj72CvxuZ6v4/C4+9BpKTnkPp/lLW71kf1fMGy+5Du1mzYw0jOo0wL7TrYybuXfhEjeq6cRLHXdl34d/n57XVr4XVl31H9jF5/mTKD5fzzLhn6JTc6eibI24ABPJeqLW8pWGEJSCo6meqeqGzvV1Vx6lqhvO8I+C4h1W1t6pmquq/w3HuEzi0F9680mieXDkrtiWuQ6FVBxhzO6z5ENb/p2FldhfDus9NMtmLw3N7n2NWbQvHaKNyv1l0JeuqqHeNVecRvN5tlO/PR9Gj6qEiMPGv0K92GfQRnUYwtutYXlzxIjsO7qj1uGCoqKrg7oV3U7izkMfOeozM1OOEB9t0NUG9YIZR57W4pmnNVK6qgn9Mgq2r4Ocvu1esbCrk3GpaSQ2drLbiLUCNdpEXaZlqFC/DIWPxzatGIymK3UXVdG3dlS6tunhe18jn99EioQWntQ9Ozvr24bdzoOIAzy4LvV9fVXnE9whfFH/B70b9jjFdx9R84Kib4eCu8MzBOYlpWgFhwUNGj2b8w5BhVQ9JbGEmq235xujI14Wq6S7qlgOpvaLjnxsycmHLEqM95JaqSih4xWjitOsdPt8aSLUcts/v87Qcts/vI6tDFolBdrn2atuLSzMu5e01b7N+9/qQfHjl21eYtWYW1w26ru7ZyN1zjBLBomftENQQaDoBYfnbZrLMsGsg55bG9sY7nHaZ+aPM/yMcqSPRV7LULKbupbkHNdEn1zwXzXdv44cFZrRMlIaa1oTX5bB3HNxB0a4iRqa7yyXdknULzeKb8eSSJ137MG/DPKbnT2f8qeOZNmxa3QeLmFZC2WrT7WlxRdMICJvzYc5ko9l/wXRv9n83FtWT1XZvBN9ztR+3bBbEN4OBF9d+jBfoNNhILYeSRyh42Qxh7Xdh+PwKEq/LYef58wCOJpSDpH2L9lw/6Hrmb5zvSrtpWdky7v3iXgZ3GMzDZzxMnDTgUjXoUjNIYHEd9dxSJ7EfEHZvhplXQutOcNmrZsUqy7H0Ogsy/gs+n17zTN/KI0bqOvN8sx6Al4mLM62EH+ZDpYvulj0lsObfMPSqRq0rXpfDzvPn0TKhJQPaDXBt45qB19CxRUem508PaojtpvJNTPl0Ch1bduSpc55quIppYnPT6lvzL6NNZQma2A4Ih/fBm1eYkQVXzvKG7o5Xyf2jWUv68z+f+N4Pn8L+bY277kEwZOQaHZ3i/ODLVieTh11b/7ERxsty2D6/j2Fpw0iMcz9ku0VCC24behsrtq1g7vq5DSqz+9Bubv3kViq1kmfGPUNq8yBHCY64wYxE8/3NhceW2A0IVVXw3i3gXwE/fxE69m9sj7xNx37mIpj3Amz/4dj3lr1pmtp9YiQR32ssSHzw3UZVlbDkFVO+EZLJx+NVOeyy/WWs270uLIvVT+g9gb4pfXlyyZMcrjxc57GHKw8zdcFUivcW839n/x892vQI/oSndIYBF8GSV80QdEtQxG5AWPiImUI//kHoG72lFWKasfdCfJLRoqnmwC4zHn/Qz2Onu61FWzOqJNiAUDQfdm9q1GRyIF6Vw67OH4QjIMTHxXPn8Dsp3lvMm9+9Wetxqsr9X91PQWkBD41+iOFpw92fdNTNcGi30eSyBEVsBoSVs2HhnyDrajj9tsb2JnZonQZnTIPV78NG5yL07RyoPBR760xn5JrW4Z6ShpcpeBmSO0Bm7ROroolX5bB9fh+tE1vTL7VfWOz9pMtPGN15NM8vf57dh3bXeMzTS5/mw7UfMmXoFC7oFeLv020kpGeZ5LIdghoUsRcQipfAe7dC99PhwsftiKJgOX0ytE43a9JWzz1o3zf2VpLLGG+eixq4aM6eLfD9R0ay2UMtIS/KYef58xieNjysC9bfPvx2yg+X8/zy5094772i93hu+XNcknEJN552Y+gnEzFDz7etMXLdlgYTWwFhzxYzoii5ozOiyENqnLFCs2Q45/cmIfvlk7DxKxj8y9gLrB0HmFnYDe02WvIqaJUnksmBeE0O27/Pz8byja6Hm9ZGZmomF/e5mDe/e5NN5UeXRVlUsogHvnqAnPQcfp/zeyRc9XDgRNMatCqoQRE7AeHIARMMDpXDlTONVo/FHUOugLRBR3MJg2OsuwhMAOtzrplkVnmk7mOrk8m9z4HUntHxr4F4TQ77x/yBywlpdTE5azLxEs9TS54CoGhnEXcsuIMebXrw+NjHQxrRdAIJSZB9PXw/98RBFJZaiY2AoGomnm1ZCpf8DdIGNrZHsU1cvBmGCkYevG23uo/3KhnjzVDajfVcTAvnwZ7NjaJbVB9ek8P2+X20SWpD35S+YbedlpzGtQOv5aP1H7Fg4wJunX8rSQlJPDPuGVo3i8BqhtnXm7puVVAbTGwEhM8fNYnkc/+3TqVFSxD0GQe5D8K5f2hsT9zT6yyIS6y/26jg72Z2s0eSycfjJTnsPH8e2WnZDZsZ7ILrBl1HavNUpiyYwq5Du3h63NOkt0qvv6AbWncyXUffvGZ6Fiz14v2AcHAXLHjYdHOMrkfPxBIco6dA1+zG9sI9Sa3h1J/UnVjevRkK55pkskfXxfCKHHbx3mKK9xaHPX8QSHJiMtOGTSMxLpFHz3w0pJnQDWLUzXBojxk8YakX7weEnRug60i48MnYS3xaIk9GLmz9FnZtqvn9Ja+aLsdh10TXryDwihy2r8QsYBiO+Qd1MTFjIl9e8SVndTsroucBzA1Pl+FmCGpDJOBPcrwfEOIS4PLXjU6JxXI8Pw4/rWGNhMqKo8nklB5RdSsYvCKHnefPI7V5Kn3aRn596RYJLSJ+jh8ZdTNsLzRLplrqxHVAEJFuIrJARFaLyCoRmeq8nioi80Sk0HlOCShzr4gUicgaEWnY9OLUXtCqo1s3LU2d9n2hbfeaF80pmgflW8z60h6nseWwVRWf30d2Wnb4hn56hQEXmxySVUGtl1BaCBXAnaraH8gBJovIAOAeYL6qZgDznX2c9y4HBgLnAc+ISP0zXxKjeCdhiT1ETCth7UKzbGog+S+bBeH7ntc4vgVBY8thbyrfROn+0oh3FzUKCc3MiKPCj+0Q1HpwHRBUtURVlzjb5cBqoAtwETDDOWwGUC2wfxEwU1UPqeo6oAhogrXPEnUyxsORfbDhq6Ov7dpkWgjDfuXZZHIg1XLYjZVY9vlN/mBEeuQSyo3K8OvMiDTbSqiTsOQQRKQHMBRYDKSpagmYoAFU9/d0AQIzf5ud12qyd5OI5ItIfllZWThctDRleowxon2B3UZLXvF8Mvl4cjrnsLRsaaPIYfv8Ptq3aE/PU7w1cS9stE4zC+gsfR0O7mlsbzxLyAFBRFoBs4FpqlrXN11Tx2SNM3FU9XlVzVbV7A4d7IxkSz00awk9zjg6H6Gywqx70Odck1+IERpLDltVyfPnMaLTiKaXPwhk1E1weC8sfaOxPfEsIQUEEUnEBIPXVfVd5+VSEUl33k8HqldD3wwETontCmwJ5fwWy49kjDcjSXasM/MOyktiIpkcSGPJYa/bs45tB7Y1zfxBIF2GmyHsPjsEtTZCGWUkwIvAalV9POCt94FqBbFrgTkBr18uIkki0hPIAHxuz2+xHENGrnku+sQkk1t3NsuGxhDNE5ozNG1o1ANCXkn41j/wPKMmwY61DVfJPckIpYUwGvgVcI6ILHUeFwCPALkiUgjkOvuo6irgLeBb4CNgsqpWhuS9xVJNu95miHLBDPNnH/YriE9obK+CJic9h+93fh9VOWyf30dayzS6tY5RTatgGHCRkX+3Kqg1Esooo/+oqqjqYFXNch7/UtXtqjpOVTOc5x0BZR5W1d6qmqmq/w7PR7BYHDLGQ+kKMxR16K8a2xtXRFsOW1XJL81nZKeRTTt/UE18ImTfAD/Mh7LvG9sbz+H9mcoWS0Op7jbqkxuzCq7RlsMu2lXEjoM7Iqpf5DmG/xrim4HvxMV6mgwulXNjr01tsdRGjzHQ70I44/bG9sQ11XLYi0oWoaoRv2uvnn8QifUPPEurDmYN8aVvwLj/geZtGtujhlN5BPZuhb2lRx/lpTXvu8AGBEvTISHJ6F7FODnpOczbMI/1e9bTs01k5wXk+fPo0qoLXVrVOCWo6TLqJlj2hpHGPn1y4/mhCkf2w4GdRx97tzoPv3ku9x/d37+9Zjst2xl5jlZpcGpvM++CB4N2xwYEi8VjBMphRzIgVGkV+aX5nN3t7Iidw7N0Hgrdcky30aibzUI6oaBqZLYDL+zHPHbV/l7l4ZptxjdzLvIdjThjt5FmjYfqC39r5zm5Yy3rhNuAYLHEPIFy2Ff0uyJi5/l+5/fsPrT75BhuWhOjJsE715kJjZnn13+8qrlb37H26GPnOmd7nQkItZGYDC1SnEdbI8r4437Kse8ldzAX+hYpUZf8twHBYvEY1XLYH6//mIqqChLiIvM3rV7/4KRKKAfS/2dmvsriZ48GhKpK2FNsLvCBF/4d68zFP1BWJC4B2p5qhjt3GwVtupmum5ou8glJjfMZg8QGBIvFg+R0zmF24WxWbV/FkA5DInKOPH8e3Vt3p1Nyp4jY9zzxiTDyRpj/R3jt57BrA+xcf2wXTnwzSOlpLvq9xkKqs53aywSAGJzrUhdN69NYLE2EUZ1GIQiLtiyKSECorKqkoLSA8T3Gh912TDHs17BslpE66ZBpWgopARf9UzqHnl+IIWxAsFg8SErzFPql9mNRySImDZkUdvvf7fiO8iPlJ2/+oJrkdnCbVdCpxk5Ms1g8SiTlsH9c/+BkzR9YasQGBIvFo0RSDtvn99GzTU86tLTy8paj2IBgsXiUSMlhH6k6wpLSJba7yHICNiBYLB4lUnLY327/lv0V+213keUEbECwWDxMtRz21v1b6z+4geT5zfoHNiBYjscGBIvFw4ztOpY4iePqf13NV8VfhcWmr8RHn7Z9SG2eGhZ7lqaDDQgWi4fpk9KHGefNICk+iUmfTOL+L+9nz2H3i8QfqTzC0rKlNn9gqREbECwWj5PVMYt3JrzDDYNuYM4Pc5j43kQWblroytaKbSs4UHHABgRLjUQ9IIjIeSKyRkSKROSeaJ/fYolFkuKTmDZ8Gm9c8AZtmrfhtk9v454v7mHXwV1B2fH5fQhCdqfsCHlqiWWiGhBEJB54GjgfGABcISIDoumDxRLLDGw/kFk/ncUtQ25h7rq5XDTnIuZtmNfg8nn+PDJTM2mTFEOLwliiRrRbCCOBIlVdq6qHgZnARVH2wWKJaRLjE7k161ZmXjiTtJZp3PHZHdzx2R1sO7CtznKHKg+xdOtSO7rIUivRDghdgE0B+5ud145BRG4SkXwRyS8rK4uacxZLLJGZmskbP32DqcOm8tmmz5g4ZyIfrv0QrWU93eVlyzlcddjmDyy1Eu2AUNNqDyfUXlV9XlWzVTW7Qwc7td5iqY2EuARuPO1G3vnZO3Q/pTv3fHEPUz6dQum+E9fU9fl9xEkcw9OGN4Knllgg2gFhM9AtYL8rsCXKPlgsTY5ebXvxynmvcHf23SwqWcTEORP5R+E/jmkt+Ep89E/tT+tmrRvRU4uXiXZAyAMyRKSniDQDLgfej7IPFkuTJD4unmsGXsPsCbPJTM3k/q/uZ9K8SWzZu4UDFQdYvm257S6y1ElU10NQ1QoRuQ2YC8QDL6nqqmj6YLE0dbqf0p0X/+tF3lrzFo8XPM7EORM5v+f5VFRV2ISypU6ivkCOqv4L+Fe0z2uxnEzESRyX97ucMV3H8MBXDzC7cDbxEs+wtGGN7ZrFw9gV0yyWJkyXVl14Lvc5Plj7AfuO7CM5MbmxXbJ4GBsQLJYmjogwofeExnbDEgNYLSOLxWKxADYgWCwWi8XBBgSLxWKxADYgWCwWi8XBBgSLxWKxADYgWCwWi8XBBgSLxWKxADYgWCwWi8VBatNO9woiUg6sCdFMe6Du1UNiz471JbJ2vORLuOxYXyJrx0u+AGSqalDStrEwU3mNqoa0AKyI5Idqw2t2rC+RteMlX8Jlx/oSWTte8qXaTrBlbJeRxWKxWAAbECwWi8XiEAsB4XmP2PCaHetLZO14yZdw2bG+RNaOl3xxZcfzSWWLxWKxRIdYaCFYLBaLJQrYgGCxWCwWwMMBQUReEpGtIrIyBBvdRGSBiKwWkVUiMtWFjeYi4hORZY6NB9z649iLF5FvROSfIdhYLyIrRGSpm6Fljo22IvKOiHznfD+nu7CR6fhQ/dgjItNc2Lnd+W5XisibItI8WBuOnamOjVXB+FFTXRORVBGZJyKFznOKSzu/cPypEpF6hxLWYuNR53daLiL/EJG2Lu086NhYKiIfi0hnN3YC3rtLRFRE2rvw5Q8iUhxQdy5w64uI/EZE1jjf859d+DIrwI/1IrLUjS8ikiUii6r/lyIy0qWdISLytfMf/0BETqnHRo3XOTd1GFX15AM4ExgGrAzBRjowzNluDXwPDAjShgCtnO1EYDGQE4JPdwBvAP8MwcZ6oH2I3+8M4EZnuxnQNkR78YAfODXIcl2AdUALZ/8t4Ncuzj8IWAm0xMyv+QTIcFvXgD8D9zjb9wB/cmmnP5AJfAZku7QxHkhwtv8Ugi+nBGxPAZ51Y8d5vRswF9hQX12sxZc/AHcF+RvXZOds57dOcvY7uvk8Ae9PB+536cvHwPnO9gXAZy7t5AFnOdvXAw/WY6PG65ybOuzZFoKqfg7sCNFGiaoucbbLgdWYC1AwNlRV9zq7ic7DVSZeRLoCPwVecFM+XDh3HGcCLwKo6mFV3RWi2XHAD6q6wUXZBKCFiCRgLuhbXNjoDyxS1f2qWgEsBCY2pGAtde0iTNDEeb7YjR1VXa2qDZ5pX4uNj53PBLAI6OrSzp6A3WQaUI/r+B8+Afw2RBtBUYudW4BHVPWQc8xWt76IiACXAW+69EWB6rv5NjSgHtdiJxP43NmeB1xaj43arnNB12HPBoRwIyI9gKGYO/xgy8Y7zcitwDxVDdqGw5OYP1GVy/LVKPCxiBSIyE0uyvcCyoCXne6rF0Qk1NXXL6cBf6TjUdVi4DFgI1AC7FbVj12cfyVwpoi0E5GWmDu0bi7sVJOmqiWOjyVAxxBshZPrgX+7LSwiD4vIJuAq4H6XNiYAxaq6zK0fDrc5XVgvNag7o2b6AmNEZLGILBSRESH4MwYoVdVCl+WnAY863+9jwL0u7awEqhfB/gVB1OPjrnNB1+GTIiCISCtgNjDtuLukBqGqlaqahbkzGykig1z4cCGwVVULgi1bA6NVdRhwPjBZRM4MsnwCppn6V1UdCuzDNCldISLNMBX4bRdlUzB3Mj2BzkCyiFwdrB1VXY3pTpkHfAQsAyrqLBRjiMh9mM/0ulsbqnqfqnZzbNzmwoeWwH24DCYB/BXoDWRhbgSmu7STAKQAOcDdwFvOnb4brsDFTU0AtwC3O9/v7TgtcBdcj/lfF2C6gA43pFCo1zk4CQKCiCRivqTXVfXdUGw53SqfAee5KD4amCAi64GZwDki8ppLP7Y4z1uBfwD1Jq+OYzOwOaCl8w4mQLjlfGCJqpa6KHsusE5Vy1T1CPAu8BM3Tqjqi6o6TFXPxDTD3d7pAZSKSDqA81xnV0SkEZFrgQuBq9TpFA6RN6inK6IWemOC9zKnLncFlohIp2CMqGqpc6NVBfyN4OtwNZuBd52uXR+m9V1nkrsmnO7KS4BZLv0AuBZTf8HcHLn6TKr6naqOV9XhmAD1Q31larnOBV2Hm3RAcO4UXgRWq+rjLm10qB7VISItMBew74K1o6r3qmpXVe2B6V75VFWDvhMWkWQRaV29jUk4BjUSS1X9wCYRyXReGgd8G6wvAYTpNKV/AAABmElEQVRyZ7URyBGRls7vNQ7TBxo0ItLRee6O+XOHcrf3PuYPjvM8JwRbISEi5wH/DUxQ1f0h2MkI2J2Au3q8QlU7qmoPpy5vxiQ0/UH6kh6wO5Eg63AA7wHnODb7YgZIuFEKPRf4TlU3u/QDTM7gLGf7HFzekATU4zjg98Cz9Rxf23Uu+DpcX9a5sR6YP3MJcART6W5wYeMMTH/7cmCp87ggSBuDgW8cGytpwAiEBtgci8tRRpj+/2XOYxVwn0s7WUC+87neA1Jc2mkJbAfahPB9PIC5OK0EXsUZMeLCzheYwLYMGBdKXQPaAfMxf+r5QKpLOxOd7UNAKTDXhY0iYFNAHW7I6KCa7Mx2vuPlwAdAFzd2jnt/PfWPMqrJl1eBFY4v7wPpLj9TM+A153MtAc5x83mAvwM3h1hnzgAKnPq3GBju0s5UzEih74FHcBQl6rBR43XOTR220hUWi8ViAZp4l5HFYrFYGo4NCBaLxWIBbECwWCwWi4MNCBaLxWIBbECwWCwWi4MNCBaLxWIBbECwWCwWi8P/A0/nwvktjhTiAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = (dga_pair_matches_df.apply(np.gradient)).apply(pd.Series).plot(\n",
    "    xticks=np.arange(20)\n",
    ")\n",
    "ax = ax.set_xticklabels(list(range(1,21)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "eswJurbzJ0Jr"
   },
   "source": [
    "It would seem the frequent character pairs for `dga` entries when considered for the entire dataset aren't illustrative. The rate of growth in `legit` classes outpaces `dga` matches at a number of steps as we increase the number of pairs to be considered.\n",
    "\n",
    "This would suggest these word pairs are perhaps too noisy to be considered as features in our model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 265
    },
    "colab_type": "code",
    "id": "5q3sjA-FIu9Z",
    "outputId": "eb899e72-90d8-41d2-9939-c633efffc3f6"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = (legit_pair_matches_df.apply(np.gradient)).apply(pd.Series).plot(\n",
    "    xticks=np.arange(20)\n",
    ")\n",
    "ax = ax.set_xticklabels(list(range(1,21)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take the top 10 most frequently occuring word pairs in the dataset for `legit` entries."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "colab_type": "code",
    "id": "5rW4imoRsbJz",
    "outputId": "4daa7cc4-09a9-4f61-8391-f9dcbd813df9"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['in', 'er', 're', 'an', 'es', 'ar', 'on', 'or', 'st', 'ne'],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "legit_character_pairs = domain_windows_count_by_class_df.loc[\"legit\", :].\\\n",
    "    sort_values(\"value\", ascending = False).head(10).reset_index()[\"variable\"].values\n",
    "\n",
    "legit_character_pairs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "56Sx8SLFyRDF"
   },
   "outputs": [],
   "source": [
    "class_matching_legit_character_pairs = data_df[(domain_windows_df[legit_character_pairs] > 0).\\\n",
    "                                                   apply(any, axis = 1)][\"class\"].value_counts()\n",
    "\n",
    "class_matching_legit_character_pairs.name = \"matching\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's plot the number of domains that contain any of the `legit_character_pairs` by class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 282
    },
    "colab_type": "code",
    "id": "xTU1AdmFyfqe",
    "outputId": "fe807fbd-96df-46a2-ff3a-3f0cef2dcc08"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2091feb46d8>"
      ]
     },
     "execution_count": 140,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame( [ total_class_counts, class_matching_legit_character_pairs] ).T.plot.barh()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's get the top most occurring `dga` character pairs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "1-iGJSTbt1Wv",
    "outputId": "78d46041-d951-47e6-b23e-8dc5500c62a6"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['yh', 'kb', 'yp', 'cu', 'yt', 'yl', 'yx', 'ws', 'qk', 'yd'],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 141,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dga_character_pairs = domain_windows_count_by_class_df.loc[\"dga\", :].sort_values(\"value\", ascending = False).head(10).reset_index()[\"variable\"].values\n",
    "dga_character_pairs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's plot the number of domains that contain any of the `dga_character_pairs` by class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "3VQe3iuPzc_E"
   },
   "outputs": [],
   "source": [
    "class_matching_dga_character_pairs = data_df[(domain_windows_df[dga_character_pairs] > 0).apply(any, axis = 1)][\"class\"].value_counts()\n",
    "class_matching_legit_character_pairs.name = \"matching\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 282
    },
    "colab_type": "code",
    "id": "yUzNY12zzgcR",
    "outputId": "1349ac13-8372-484a-9d1e-d722ad95efa0"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2091ff2b160>"
      ]
     },
     "execution_count": 143,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame( [ total_class_counts, class_matching_dga_character_pairs] ).T.plot.barh()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Observations on Character Pair Windows\n",
    "\n",
    "Taking a sample windowed character pairs from `legit` domains ordered by their frequency of appearance as a feature may provide some discriminatory value. The `dga` character pairs are too noisy, probably a result the `words` being generated."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "qXODOLJoQqOv"
   },
   "source": [
    "# Frequency Analysis\n",
    "\n",
    "If we assigned classes based on their frequency what level of accuracy would we achieve?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "BT8a_sjmTx0N"
   },
   "outputs": [],
   "source": [
    "def get_probabilities(df, column):\n",
    "    return (pd.DataFrame(df[column].value_counts()) / len(df)).to_dict()[column]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "J-RhfL4gUbzU",
    "outputId": "4983a7a3-b9eb-425c-8fee-072da1abc49b"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'legit': 0.6067604498006361, 'dga': 0.39323955019936385}"
      ]
     },
     "execution_count": 145,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "class_probabilities = get_probabilities(data_df, \"class\")\n",
    "class_probabilities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "qApjsvlxBxwp"
   },
   "outputs": [],
   "source": [
    "def random_prob_assignment(probabilities_dict):\n",
    "    return np.random.choice(list(probabilities_dict.keys()), p = list(probabilities_dict.values()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "tPTTCaoit9MN"
   },
   "outputs": [],
   "source": [
    "def probability_assignment_accuracy(df, column, probabilities_dict):\n",
    "    assigned_values = df[column].apply(lambda x: random_prob_assignment(probabilities_dict))\n",
    "    return len(df[assigned_values == df[column]]) / len(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "yAAHdy47UhUU"
   },
   "outputs": [],
   "source": [
    "number_of_trials = 10\n",
    "class_probability_assignment_trials = pd.Series([probability_assignment_accuracy(data_df, \"class\", class_probabilities) for x in range(number_of_trials)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 285
    },
    "colab_type": "code",
    "id": "Qx11kXpMUm2z",
    "outputId": "d869a1fb-2a5b-4dfa-bbda-a7fd79efb25b"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2091ff8ca90>"
      ]
     },
     "execution_count": 149,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "class_probability_assignment_trials.plot.kde()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 170
    },
    "colab_type": "code",
    "id": "mauoHBLQUpR0",
    "outputId": "b9a18e4c-28ac-4f73-8791-c6bdf0878f69"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    10.000000\n",
       "mean      0.522942\n",
       "std       0.001389\n",
       "min       0.521086\n",
       "25%       0.521881\n",
       "50%       0.522946\n",
       "75%       0.523845\n",
       "max       0.525499\n",
       "dtype: float64"
      ]
     },
     "execution_count": 150,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "class_probability_assignment_trials.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "7QQ6jN5y6hXh"
   },
   "source": [
    "Assigning the class based on frequency seems to converge on 52% accuracy, so a favourable coin toss."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 85
    },
    "colab_type": "code",
    "id": "7iHANaEmj2ka",
    "outputId": "b8bfb406-b6f9-4755-b233-bec91f8099bc"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'legit': 0.6067604498006361,\n",
       " 'cryptolocker': 0.2562534533996386,\n",
       " 'newgoz': 0.08212744351358213,\n",
       " 'goz': 0.054858653286143096}"
      ]
     },
     "execution_count": 151,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "subclass_probabilities = get_probabilities(data_df, \"subclass\")\n",
    "subclass_probabilities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "2ge7cVYHucvh"
   },
   "outputs": [],
   "source": [
    "number_of_trials = 10\n",
    "subclass_probability_assignment_trials = pd.Series([probability_assignment_accuracy(data_df, \"subclass\", subclass_probabilities) for x in range(number_of_trials)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 282
    },
    "colab_type": "code",
    "id": "0EP5I0nLCBA7",
    "outputId": "f6cf3110-8514-4d5c-b022-1ddfd2c00cbe"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2094cebc828>"
      ]
     },
     "execution_count": 153,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "subclass_probability_assignment_trials.plot.kde()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 170
    },
    "colab_type": "code",
    "id": "c6AyJ5qkC2BF",
    "outputId": "5a201237-c79b-4dab-ddcd-4af58e2e340a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    10.000000\n",
       "mean      0.444362\n",
       "std       0.001168\n",
       "min       0.442476\n",
       "25%       0.444122\n",
       "50%       0.444279\n",
       "75%       0.444718\n",
       "max       0.446687\n",
       "dtype: float64"
      ]
     },
     "execution_count": 154,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "subclass_probability_assignment_trials.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "m7rRaP2k1gaT"
   },
   "source": [
    "## Observations on Frequency Analysis\n",
    "\n",
    "If we developed a model purely on frequency of class we'd expect an accuaracy of ~52%\n",
    "\n",
    "We could certainly do better with the application of appropriate models."
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "Copy of Phishing Urls - EDA",
   "provenance": [],
   "toc_visible": true
  },
  "kernelspec": {
   "display_name": "domain_generation_algorithm_classifier",
   "language": "python",
   "name": "domain_generation_algorithm_classifier"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}