Emotion Classification Using Naive Bayes

This repository contains a Python project that builds an emotion classification model from text data using a Multinomial Naive Bayes classifier. The workflow includes loading and preprocessing the data, feature extraction with TF-IDF, model training, evaluation, and making predictions on new text.

Table of Contents

• Overview
• Data Preparation
• Exploratory Data Analysis
• Text Preprocessing
• Feature Extraction
• Model Training and Evaluation
• Example Prediction
• Dependencies
• Usage

Overview

The project performs emotion classification by:

• Reading a dataset containing text and corresponding emotion labels.
• Replacing numerical labels with emotion names.
• Preprocessing the text to remove stopwords and perform lemmatization.
• Converting text into numerical features using TF-IDF.
• Splitting the data into training and testing sets.
• Training a Naive Bayes classifier.
• Evaluating the model's performance and making a prediction on new input.

Data Preparation

1. Data Loading:
   The dataset is loaded from a CSV file (emotions.csv) using pandas.

2. Label Mapping:
   The code replaces numerical labels (0-5) with corresponding emotion names:

   • 0 → "sadness"
   • 1 → "joy"
   • 2 → "love"
   • 3 → "anger"
   • 4 → "fear"
   • 5 → "surprise"

3. Subsetting Data:
   For efficiency, only the first 5000 rows are used for further analysis.

Exploratory Data Analysis

• Dataset Shape & Missing Values:
  The code prints the shape of the dataset and checks for any missing values.

• Label Distribution:
  It prints the frequency distribution of the emotion labels to understand class balance.

• Data Information:
  General information and summary statistics about the dataset are displayed.

Text Preprocessing

• SpaCy Initialization:
  The SpaCy English model (en_core_web_sm) is loaded to perform NLP tasks.

• Preprocessing Function:
  A function preprocess_text is defined to:

  • Convert text to lowercase.
  • Tokenize the text.
  • Remove stopwords.
  • Keep only alphabetic tokens.
  • Apply lemmatization.

• Applying Preprocessing:
  The function is applied to the text column in the dataset, and a new column cleaned_text is created.

Feature Extraction

• TF-IDF Vectorization:
  The TfidfVectorizer is initialized with a maximum of 1000 features.

  • The cleaned text data is transformed into a TF-IDF feature matrix.

• Train-Test Split:
  The feature matrix (X) and target labels (y) are split into training and testing sets using an 80/20 ratio.

Model Training and Evaluation

• Training:
  A Multinomial Naive Bayes classifier is instantiated and trained on the training data.

• Prediction and Evaluation:
  The classifier predicts emotions on the test set.

  • The model's accuracy is calculated.
  • A classification report is generated, showing precision, recall, and F1-score for each emotion.

Example Prediction

An example is provided where a new text ("He seemed mesmerized") is:

• Preprocessed using the same preprocess_text function.
• Transformed using the trained TF-IDF vectorizer.
• Classified using the trained Naive Bayes classifier, outputting the predicted emotion.

Dependencies

• Python Libraries:

  • numpy
  • pandas
  • scikit-learn
  • spacy

• SpaCy Model:

  • en_core_web_sm (Make sure to download it via python -m spacy download en_core_web_sm)

Usage

1. Clone the Repository
   Clone this repository to your local machine.

2. Install Dependencies
   Install the required libraries using pip:

   pip install numpy pandas scikit-learn spacy
   python -m spacy download en_core_web_sm