Movies recommender system using the Singular Value Decomposition method
Herby is an example of recommender system applied to a dataset of ratings for movies.
Content-based filtering using K-nearest neighbor (KNN) classifier / similarity methods.
The movies matrix contains 1682 movies produced from 1930 till 1998. It contains the variables Y and R:
- R: a (num_movies x num_users)-matrix of size 1682 x 943 containing the ratings (from 1 to 5) given by 943 users for 1682 movies (of course not all the movies are rated)
- Y: a (num_movies x num_users)-matrix of size 1682 x 943. This matrix has only values of 0 or 1.
NOTE: Each cell y_{ij} is 1 if the user i has rated the movie j. The cell y_{ij} is 0 if the user i did not rate the movie j. movie_ids.text: contains the movie ids and the name of the movies.
Credits for the dataset goes to Prof. Dr. Wolfgang Konen at TH-Cologne, Germany.
As a use case the movies recommended for the User 12 in dataset has been listed.
- 'Using Singular Value Decomposition to Build a Recommender System', Adrian Tam, October 29, 2021, link