Music Recommender System using Non-Negative Matrix Factorization

Recommender systems play a crucial role in music streaming services like Spotify to help users discover music that aligns with their preferences. The goal of this research is to create a music recommendation system that utilizes collaborative filtering with Non-Negative Matrix Factorization (NMF). Matrix Factorization decomposes user-music interactions into smaller matrices to predict user preferences for songs they have not listened to yet. This method will be compared with SVD to see if NMF is better at recommending music that match their likes and preferences. NMF imposes a non-negativity constraint on the decomposed matrices, making the resulting factors easier to interpret. Performance evaluation is conducted using RMSE (Root Mean Squared Error) and MAE (Mean Absolute Error) metrics to find out how good the method provides recommendations. Based on the test results, the NMF method with an RMSE value of 0.4727 and MAE value 0.3693, indicating its effectiveness in providing music recommendations.