Book Recommendation System

My Contribution

• Processed the Book-Crossing Books, Ratings, and Users datasets.
• Cleaned book metadata and user data, including invalid publication years and missing ages.
• Separated explicit ratings from implicit feedback and normalized user age data.
• Merged the datasets and filtered users and books with at least 50 ratings to reduce sparsity.
• Built User-Based Collaborative Filtering using cosine similarity between users.
• Implemented personalized recommendations with SVD matrix factorization using Surprise.
• Developed a Streamlit dashboard for book exploration, personalized recommendations, and model insights.
• Evaluated and compared both recommendation approaches using RMSE, with SVD achieving the lower error.