Tutorial/9_Genetic_Algorithm_Overview.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Objective functions can optionally take in step, budget, and generations.\n",
    "\n",
    "step - The same objective function will be run for #evaluation_early_stop_steps, the current step will be passed into the function as an interger. (This is useful for getting a single fold of cross validation for example).\n",
    "\n",
    "budget - A parameter that varies over the course of the generations. Gets passed into the objective function as a float between 0 and 1. If the budget of the previous evaluation is less than the current budget, it will get re-evaluated. Useful for using smaller datasets earlier in training.\n",
    "\n",
    "generations - an int corresponding to the current generation number.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Generation: 100%|██████████| 100/100 [02:15<00:00,  1.35s/it]\n"
     ]
    }
   ],
   "source": [
    "#knapsack problem\n",
    "import numpy as np\n",
    "import tpot\n",
    "import random\n",
    "import matplotlib.pyplot as plt\n",
    "from dask.distributed import Client, LocalCluster\n",
    "\n",
    "class SubsetSelector(tpot.individual.BaseIndividual):\n",
    "    def __init__(   self,\n",
    "                    values,\n",
    "                    initial_set = None,\n",
    "                    k=1, #step size for shuffling\n",
    "                ):\n",
    "\n",
    "        if isinstance(values, int):\n",
    "            self.values = set(range(0,values))\n",
    "        else:\n",
    "            self.values = set(values)\n",
    "\n",
    "\n",
    "        if initial_set is None:\n",
    "            self.subsets = set(random.choices(values, k=k))\n",
    "        else:\n",
    "            self.subsets = set(initial_set)\n",
    "\n",
    "        self.k = k\n",
    "\n",
    "        self.mutation_list = [self._mutate_add, self._mutate_remove]\n",
    "        self.crossover_list = [self._crossover_swap]\n",
    "        \n",
    "\n",
    "    def mutate(self, rng=None):\n",
    "        mutation_list_copy = self.mutation_list.copy()\n",
    "        random.shuffle(mutation_list_copy)\n",
    "        for func in mutation_list_copy:\n",
    "            if func():\n",
    "                return True\n",
    "        return False\n",
    "\n",
    "    def crossover(self, ind2, rng=None):\n",
    "        crossover_list_copy = self.crossover_list.copy()\n",
    "        random.shuffle(crossover_list_copy)\n",
    "        for func in crossover_list_copy:\n",
    "            if func(ind2):\n",
    "                return True\n",
    "        return False\n",
    "\n",
    "    def _mutate_add(self,):\n",
    "        not_included = list(self.values.difference(self.subsets))\n",
    "        if len(not_included) > 1:\n",
    "            self.subsets.update(random.sample(not_included, k=min(self.k, len(not_included))))\n",
    "            return True\n",
    "        else:\n",
    "            return False\n",
    "\n",
    "    def _mutate_remove(self,):\n",
    "        if len(self.subsets) > 1:\n",
    "            self.subsets = self.subsets - set(random.sample(list(self.subsets), k=min(self.k, len(self.subsets)-1) ))\n",
    "\n",
    "    def _crossover_swap(self, ss2):\n",
    "        diffs = self.subsets.symmetric_difference(ss2.subsets)\n",
    "\n",
    "        if len(diffs) == 0:\n",
    "            return False\n",
    "        for v in diffs:\n",
    "            self.subsets.discard(v)\n",
    "            ss2.subsets.discard(v)\n",
    "            random.choice([self.subsets, ss2.subsets]).add(v)\n",
    "        \n",
    "        return True\n",
    "\n",
    "    def unique_id(self):\n",
    "        return str(tuple(sorted(self.subsets)))\n",
    "\n",
    "def individual_generator():\n",
    "    while True:\n",
    "        yield SubsetSelector(values=np.arange(len(values)))\n",
    "\n",
    "\n",
    "values = np.random.randint(200,size=100)\n",
    "weights = np.random.random(200)*10\n",
    "max_weight = 50\n",
    "\n",
    "def simple_objective(ind, **kwargs):\n",
    "    subset = np.array(list(ind.subsets))\n",
    "    if len(subset) == 0:\n",
    "        return 0, 0\n",
    "\n",
    "    total_weight = np.sum(weights[subset])\n",
    "    total_value = np.sum(values[subset])\n",
    "\n",
    "    if total_weight > max_weight:\n",
    "        total_value = 0\n",
    "\n",
    "    return total_value, total_weight\n",
    "\n",
    "objective_names = [\"Value\", \"Weight\"]\n",
    "objective_function_weights = [1,-1]\n",
    "\n",
    "\n",
    "\n",
    "evolver = tpot.evolvers.BaseEvolver(   individual_generator=individual_generator(), \n",
    "                                objective_functions=[simple_objective],\n",
    "                                objective_function_weights = objective_function_weights,\n",
    "                                bigger_is_better = True,\n",
    "                                population_size= 100,\n",
    "                                objective_names = objective_names,\n",
    "                                generations= 100,\n",
    "                                n_jobs=32,\n",
    "                                verbose = 1,\n",
    "\n",
    ")\n",
    "\n",
    "evolver.optimize()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "best subset {3, 7, 8, 10, 13, 22, 31, 42, 43, 51, 57, 60, 64, 67, 68, 76, 80, 83, 97, 98}\n",
      "Best value 2925.0, weight 49.597868834152706\n",
      "\n",
      "All results\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>Selected Index</th>\n",
       "      <th>Value</th>\n",
       "      <th>Weight</th>\n",
       "      <th>Parents</th>\n",
       "      <th>Variation_Function</th>\n",
       "      <th>Individual</th>\n",
       "      <th>Generation</th>\n",
       "      <th>Submitted Timestamp</th>\n",
       "      <th>Completed Timestamp</th>\n",
       "      <th>Eval Error</th>\n",
       "      <th>Pareto_Front</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>(70,)</td>\n",
       "      <td>44.0</td>\n",
       "      <td>0.834758</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>&lt;__main__.SubsetSelector object at 0x7b8da0987...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>(42,)</td>\n",
       "      <td>147.0</td>\n",
       "      <td>3.091616</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>&lt;__main__.SubsetSelector object at 0x7b8da0987...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>(90,)</td>\n",
       "      <td>95.0</td>\n",
       "      <td>6.653284</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>&lt;__main__.SubsetSelector object at 0x7b8da0985...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>(94,)</td>\n",
       "      <td>159.0</td>\n",
       "      <td>7.523552</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>&lt;__main__.SubsetSelector object at 0x7b8da0985...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>(97,)</td>\n",
       "      <td>184.0</td>\n",
       "      <td>2.483618</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>&lt;__main__.SubsetSelector object at 0x7b8da0987...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9995</th>\n",
       "      <td>(61, 71, 99)</td>\n",
       "      <td>307.0</td>\n",
       "      <td>17.213843</td>\n",
       "      <td>((61,), (61,))</td>\n",
       "      <td>ind_mutate</td>\n",
       "      <td>&lt;__main__.SubsetSelector object at 0x7b8cd5553...</td>\n",
       "      <td>99.0</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9996</th>\n",
       "      <td>(7, 8, 22, 36, 43, 51, 61, 67, 68, 75, 80, 98)</td>\n",
       "      <td>1457.0</td>\n",
       "      <td>20.259947</td>\n",
       "      <td>((7, 8, 22, 36, 43, 51, 61, 67, 68, 80, 98), (...</td>\n",
       "      <td>ind_mutate</td>\n",
       "      <td>&lt;__main__.SubsetSelector object at 0x7b8cd5553...</td>\n",
       "      <td>99.0</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9997</th>\n",
       "      <td>(7, 8, 20, 22, 43, 61, 67, 98)</td>\n",
       "      <td>997.0</td>\n",
       "      <td>14.617154</td>\n",
       "      <td>((7, 8, 22, 43, 61, 98), (7, 8, 22, 43, 61, 98))</td>\n",
       "      <td>ind_mutate</td>\n",
       "      <td>&lt;__main__.SubsetSelector object at 0x7b8cd5553...</td>\n",
       "      <td>99.0</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9998</th>\n",
       "      <td>(7, 8, 22, 25, 36, 43, 51, 67, 78, 98)</td>\n",
       "      <td>1167.0</td>\n",
       "      <td>10.815193</td>\n",
       "      <td>((7, 8, 22, 36, 43, 51, 67, 78, 98), (7, 8, 22...</td>\n",
       "      <td>ind_mutate</td>\n",
       "      <td>&lt;__main__.SubsetSelector object at 0x7b8cd5553...</td>\n",
       "      <td>99.0</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9999</th>\n",
       "      <td>(0, 7, 8, 22, 36, 43, 51, 67, 68, 98)</td>\n",
       "      <td>1318.0</td>\n",
       "      <td>15.988821</td>\n",
       "      <td>((7, 8, 22, 36, 43, 51, 67, 68, 98), (7, 8, 22...</td>\n",
       "      <td>ind_mutate</td>\n",
       "      <td>&lt;__main__.SubsetSelector object at 0x7b8cd5553...</td>\n",
       "      <td>99.0</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>1.727561e+09</td>\n",
       "      <td>None</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10000 rows × 11 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                      Selected Index   Value     Weight  \\\n",
       "0                                              (70,)    44.0   0.834758   \n",
       "1                                              (42,)   147.0   3.091616   \n",
       "2                                              (90,)    95.0   6.653284   \n",
       "3                                              (94,)   159.0   7.523552   \n",
       "4                                              (97,)   184.0   2.483618   \n",
       "...                                              ...     ...        ...   \n",
       "9995                                    (61, 71, 99)   307.0  17.213843   \n",
       "9996  (7, 8, 22, 36, 43, 51, 61, 67, 68, 75, 80, 98)  1457.0  20.259947   \n",
       "9997                  (7, 8, 20, 22, 43, 61, 67, 98)   997.0  14.617154   \n",
       "9998          (7, 8, 22, 25, 36, 43, 51, 67, 78, 98)  1167.0  10.815193   \n",
       "9999           (0, 7, 8, 22, 36, 43, 51, 67, 68, 98)  1318.0  15.988821   \n",
       "\n",
       "                                                Parents Variation_Function  \\\n",
       "0                                                   NaN                NaN   \n",
       "1                                                   NaN                NaN   \n",
       "2                                                   NaN                NaN   \n",
       "3                                                   NaN                NaN   \n",
       "4                                                   NaN                NaN   \n",
       "...                                                 ...                ...   \n",
       "9995                                     ((61,), (61,))         ind_mutate   \n",
       "9996  ((7, 8, 22, 36, 43, 51, 61, 67, 68, 80, 98), (...         ind_mutate   \n",
       "9997   ((7, 8, 22, 43, 61, 98), (7, 8, 22, 43, 61, 98))         ind_mutate   \n",
       "9998  ((7, 8, 22, 36, 43, 51, 67, 78, 98), (7, 8, 22...         ind_mutate   \n",
       "9999  ((7, 8, 22, 36, 43, 51, 67, 68, 98), (7, 8, 22...         ind_mutate   \n",
       "\n",
       "                                             Individual  Generation  \\\n",
       "0     <__main__.SubsetSelector object at 0x7b8da0987...         0.0   \n",
       "1     <__main__.SubsetSelector object at 0x7b8da0987...         0.0   \n",
       "2     <__main__.SubsetSelector object at 0x7b8da0985...         0.0   \n",
       "3     <__main__.SubsetSelector object at 0x7b8da0985...         0.0   \n",
       "4     <__main__.SubsetSelector object at 0x7b8da0987...         0.0   \n",
       "...                                                 ...         ...   \n",
       "9995  <__main__.SubsetSelector object at 0x7b8cd5553...        99.0   \n",
       "9996  <__main__.SubsetSelector object at 0x7b8cd5553...        99.0   \n",
       "9997  <__main__.SubsetSelector object at 0x7b8cd5553...        99.0   \n",
       "9998  <__main__.SubsetSelector object at 0x7b8cd5553...        99.0   \n",
       "9999  <__main__.SubsetSelector object at 0x7b8cd5553...        99.0   \n",
       "\n",
       "      Submitted Timestamp  Completed Timestamp Eval Error  Pareto_Front  \n",
       "0            1.727561e+09         1.727561e+09       None           NaN  \n",
       "1            1.727561e+09         1.727561e+09       None           NaN  \n",
       "2            1.727561e+09         1.727561e+09       None           NaN  \n",
       "3            1.727561e+09         1.727561e+09       None           NaN  \n",
       "4            1.727561e+09         1.727561e+09       None           NaN  \n",
       "...                   ...                  ...        ...           ...  \n",
       "9995         1.727561e+09         1.727561e+09       None           NaN  \n",
       "9996         1.727561e+09         1.727561e+09       None           NaN  \n",
       "9997         1.727561e+09         1.727561e+09       None           NaN  \n",
       "9998         1.727561e+09         1.727561e+09       None           NaN  \n",
       "9999         1.727561e+09         1.727561e+09       None           NaN  \n",
       "\n",
       "[10000 rows x 11 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "final_population_results = evolver.population.evaluated_individuals\n",
    "final_population_results.reset_index(inplace=True)\n",
    "final_population_results = final_population_results.rename(columns = {'index':'Selected Index'})\n",
    "\n",
    "best_idx = final_population_results[\"Value\"].idxmax()\n",
    "best_individual = final_population_results.loc[best_idx]['Individual']\n",
    "print(\"best subset\", best_individual.subsets)\n",
    "print(\"Best value {0}, weight {1}\".format(final_population_results.loc[best_idx, \"Value\"],final_population_results.loc[best_idx, \"Weight\"]))\n",
    "print()\n",
    "\n",
    "print(\"All results\")\n",
    "final_population_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x400 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from scipy.stats import binned_statistic_2d\n",
    "\n",
    "y = final_population_results[\"Value\"]\n",
    "x = final_population_results[\"Weight\"]\n",
    "c = final_population_results[\"Generation\"]\n",
    "\n",
    "x_bins = np.linspace(0, 100, 100)\n",
    "y_bins = np.linspace(0, 3000, 100)\n",
    "\n",
    "ret = binned_statistic_2d(x, y, c, statistic=np.mean, bins=[x_bins, y_bins])\n",
    "\n",
    "fig, ax1 = plt.subplots(1, 1, figsize=(12, 4))\n",
    "\n",
    "im = ax1.imshow(ret.statistic.T, origin='lower', extent=(0,100,0,3000), vmin=0, vmax=100, aspect=.03)\n",
    "ax1.set_xlabel(\"Weight\")\n",
    "ax1.set_ylabel(\"Value\")\n",
    "ax1.set_title(\"Binned Average Generation\")\n",
    "\n",
    "cbar = fig.colorbar(im,)\n",
    "cbar.set_label('Generation')\n",
    "plt.tight_layout()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "tpot_dev",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.14"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "7fe1fe9ef32cd5efd76326a08046147513534f0dd2318301a1a96ae9071c1c4e"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}