Who would you listen to?: A Frequent Itemset Mining and Recommender System approach on the Million Song Dataset
Big Data and Cloud Computing mini-project by Justine Buño, Elijah Justin Medina & Raphael Mari Ongleo
With the MillionSong dataset, the listening patterns of select users were analyzed using FIM analysis. The artists that are commonly listened to by the same user is analyzed and different association rules were extracted. After that, a recommender system was built by recommending new artists to the users. This model was analyzed by comparing the results of the recommendation with the similarity of the artists in genre to gain a deeper understanding on how the model creates its recommendation. It was found that genres have minimal to no effect on the recommendations. To better understand the listening patterns of users, other information should be explored.
Report and analysis
If you have any questions regarding this study or wish to have a copy of the report, please send me a message via e-mail or LinkedIn. The code used for this analysis is available here.