Dataset. MovieLens Dataset was used for the project. (link: https://grouplens.org/datasets/movielens/100k/). MovieLens 100K movie ratings. 14848 ratings from 50 users on 9742 movies was used for recommendation website.
For Recommendation system User Based Collaborative Filtering was applied.
Collaborative filtering is one of the approaches to predicting behavior (more precisely, choice) within the framework of a recommendation system. Algorithm based on the so-called utility matrix, in the columns of which are the items of recommendations (items), and in the rows - users (users). Each cell reflects the attitude of the user U to the subject I. These relations - ratings are used to determine the similarity of items based on the proximity of the ratings received by these items from the same user. Users are considered similar if their vectors are close according to some measure of mathematical distance. In the next step, a recommendation for the user is drawn up based on the preferences of similar users.