The aim of this project was to create a movie recommendation system based on a dataset containing the ratings of 6040 netflix users for 3706 movies.
Our training data consisted of 3 datasets, containing data about movies , users, as well as user ratings for those movies. The movies dataset contained the year and the title for each movie; the users dataset contained the gender, age, and a number encoding their profession; most importantly, the ratings dataset contained the ratings some users gave to certain movies.
What we, essentially tried to achieve was to make use of the available data predict future ratings as accurately as possible. Our approach was based on 2 directions, according to which we split our team:
- Collaborative Filtering - Cătălin Lupău
- Latent Factor Decomposition - Pietro Vigilanza
For the full report of this project see this document.