Sentiment Analysis Project

Overview

This project implements a sentiment analysis system using various machine learning models. It processes textual data, extracts features, trains multiple classification models, and evaluates their performance in predicting sentiment.

Features

• Data processing and feature extraction from raw text
• Implementation of multiple machine learning models:
  • Logistic Regression
  • Random Forest
  • Support Vector Machine (SVM)
  • Naive Bayes
• Model evaluation with various performance metrics
• Visualization of results including confusion matrices

Installation

1. Clone this repository:

   git clone https://github.com/reetmitra/Twitter-Sentiment-Analysis.git
   cd Twitter-Sentiment-Analysis

2. Create a venv (Optional):

   python -m venv venv
   source venv/bin/activate

3. Install the required packages:

   pip install -r requirements.txt

Usage

1. Data Processing:

• Place your raw data in the data/raw/ directory
• Run the data processing script:

  python src/data_processing.py
  

2. Feature Extraction:

• After data processing, run the feature extraction:

  python src/feature_extraction.py
  

3. Model Training:

• Train the models using:

  python src/model_training.py
  

4. Evaluation:

• Evaluate the trained models:

  python src/evaluation.py
  

Results

The performance of each model is evaluated using the following metrics:

• Accuracy
• Precision
• Recall
• F1-score

Confusion matrices for each model are saved in the figures/ directory.

Models

1. Logistic Regression: A linear model for binary classification.
2. Random Forest: An ensemble learning method using multiple decision trees.
3. Support Vector Machine (SVM): A powerful algorithm for both linear and non-linear classification.
4. Naive Bayes: A probabilistic classifier based on applying Bayes' theorem.

Future Work

• Implement deep learning models (e.g., LSTM, BERT)
• Experiment with different feature extraction techniques
• Develop a web interface for real-time sentiment analysis
• Explore multi-class sentiment classification

Contributing

Contributions to this project are welcome! Please fork the repository and submit a pull request with your proposed changes.

Contact

Project Link: https://github.com/reetmitra/Twitter-Sentiment-Analysis