Amazon Product Recommender System

This project uses collaborative filtering and matrix factorization techniques to recommend Amazon products to users based on their review history and rating behavior. The goal is to improve user satisfaction and increase conversion rates by delivering relevant, personalized suggestions.

Key Visualizations

Model Performance

The following table shows key metrics for the final models evaluated:

Business Takeaways

• User-based CF shows high recall, potentially increasing product discovery effectiveness.
• SVD offered consistent performance and handles sparse data efficiently.
• Future improvements include bias mitigation and cold-start solutions.