Collaborative Filtering Movie Recommender

Pick a user to get two Top-5 lists — one from User-Based CF (similar users) and one from Item-Based CF (similar movies). Add a movie to predict that user's rating both ways.

Loading MovieLens 100K...

Similarity: cosine on co-rated items, weighted by the number of co-rated items (min(n, 50) / 50). ★ = predicted rating on the 1–5 scale; lists are ranked by the evidence-weighted score Σ sim × rating.