Collaborative Filtering Movie Recommendation

Collaborative Filtering Movie Recommendation algorithm.

Tools

Movie Recommendation task by using:

• Item-based Collaborative Filtering
• Matrix Mean Normalization
• Adjusted Cosine Distance
• k-Nearest Neighbors selection for computed similarities (default: k = 0 (disabled))
• Item cold-start:
  1. Uses item average rating as the <user, item> prediction
  2. If the item average doesn't exists, uses global items average rating as the <user, item> prediction

Evaluation Metric

• Accuracy metric: RMSE (Root Mean Squared Error)