Movie Recommender System & A/B Testing Framework

A production-grade movie recommendation system demonstrating Collaborative Filtering (SVD), Popularity baselines, and a rigorous A/B Testing experimentation framework.

Features

• Data Pipeline: Automated download an processing of MovieLens 100k data.
• Recommendations:
  • Popularity Baseline: Fallback for cold-start users.
  • SVD (Matrix Factorization): Collaborative filtering for personalized suggestions.
• A/B Testing Simulator:
  • Deterministic user bucketing (Hashing).
  • Simulation engine using held-out test data ("Future Ground Truth") to model user behavior.
  • Statistical analysis (Z-Test) to measure lift.
• Interactive Dashboard: Streamlit app for exploring recommendations and running live experiment simulations.

Tech Stack

• Python 3.11 (Pyenv + Pip)
• ML: Scikit-Learn (TruncatedSVD), NumPy, Pandas
• Stats: SciPy, Statsmodels
• App: Streamlit, Plotly

Installation

1. Clone the repository
2. Setup Python Environment (Recommended: pyenv) Ensure you have Python 3.11+ installed.

   pyenv local 3.11.14  # or your preferred version
   python -m venv .venv
   source .venv/bin/activate

3. Install dependencies

   pip install -r requirements.txt

Quick Start

1. Data Setup

Download and process the MovieLens dataset:

python src/data/loader.py

2. Run the Dashboard

Launch the interactive application:

streamlit run src/app/dashboard.py

Open http://localhost:8501 in your browser.

• User View: Login as any user ID (e.g., 42) to see your assigned group and recommendations.
• Admin View: Go to 'Admin Dashboard' to run a full A/B test simulation and see the statistical results.

Docker Support

Build and run the application in a container:

docker build -t movie-recommender .
docker run -p 8501:8501 movie-recommender

Development & Testing

Run the test suite to verify model mechanics and experiment logic:

pip install pytest
python -m pytest tests/

Running the Experiment Script

You can also run the A/B test simulation directly from the CLI:

export PYTHONPATH=$PYTHONPATH:.
python src/experiment/ab_test.py

Project Structure

├── data/                   # Raw and processed data
├── notebooks/              # EDA and Prototyping
├── src/
│   ├── app/                # Streamlit Dashboard
│   ├── data/               # Data Loaders
│   ├── experiment/         # A/B Testing Logic & Stats
│   └── models/             # Recommender Classes
└── requirements.txt        # Dependencies