This project involved exploratory analysis of a movie ratings dataset, with the goal of predicting the rating a user would give to a movie based on ratings from other users.
Two manually built linear regression models were used to predict ratings: one based solely on users and their ratings, and another incorporating movie metadata as well.
The FunkSVD model was also implemented.
The methodology included selecting the best hyperparameters through cross-validation, as well as choosing the final model based on RMSE and MAE metrics. Additionally, Precision@k, Recall@k, F1@k, and MRR@k metrics were used to evaluate the recommendation models’ performance.