Kaggle Assignment 3 – Sentiment Analysis

This repository contains a refactored solution for Kaggle Assignment 3 of the Machine Learning Practice (MLP) course, IIT Madras.

The objective of this assignment is to perform multi-class sentiment analysis on movie review phrases using classical machine learning models and TF-IDF-based text features.

---

Project Structure

The project has been refactored from a single Jupyter Notebook into a modular structure:

• src/: Contains Python modules for data preprocessing (data_preprocessing.py), model training (model_training.py), and configuration (config.py).
• scripts/: Contains executable scripts for training (train.py) and prediction (predict.py).
• notebook/: Contains the refactored Jupyter Notebook, which now serves as a high-level interface and for visualizations.
• data/: (Not included in repo) This is where the Kaggle dataset files should be placed.
• requirements.txt: Lists the Python dependencies for the project.

---

Installation

1. Clone the repository:

   git clone https://github.com/your-username/your-repo-name.git
   cd your-repo-name

2. Install the required libraries:

   pip install -r requirements.txt

3. Download the dataset from the Kaggle competition page and place the train.csv, test.csv, and sample_submission.csv files into a /kaggle/input/mlp-term-3-2025-kaggle-assignment-3 directory.

---

Usage

Training the Model

To train the sentiment analysis model, run the train.py script:

python scripts/train.py

This will preprocess the training data, train the final model (Character n-gram TF-IDF + Logistic Regression), and save the trained model to model.joblib.

Generating Predictions

To generate predictions on the test set, run the predict.py script:

python scripts/predict.py

This will load the trained model, make predictions on the test data, and create a submission.csv file in the required Kaggle format.

---

Author

Kashyap Rahul
IIT Madras – BS Degree Programme
Machine Learning Practice