Collaborative Filtering Approach for Movie Recommendations

This paper describes the development of a movies recommendation system by considering the history of the movie watched by the users through the collaborative filtering approach. This analysis are based on two datasets: the movies dataset and the ratings dataset with a total of 10,329 movie titles. Next, the matrix of the movie genre is built, and the search matrix is produced with 18 genres identified. The other matrices such as user and movie similarities matrices were created to ease the analysis of filtering. The visualization results showed that the movie Pulp Fiction has the greatest number of viewers, which is 325 views. The item-based collaborative filtering approach is also used to predict the movie to be recommended and the result revealed that two movies; Get Shorty (1995) and Bad Boys (1995) are highly recommended by the users. The item-based collaborative filtering helps the users to filter the movies according to their preference.