Here you will build recommendation models for a Music Platform and try to predict what ratings customers will assign to songs. You can use all the extra information provided, such as customers's previous rating scores, his/her list of songs saved for listening in future, labels assigned to songs etc.
Dataset contains ~1.3million ratings, split into about 700k training and 600k test ratings. There are about 14k distinct customers and 10k distinct songs. This dataset is derived from a real world scenario; so take care of sanitizing/handling real data.
Use the data in creative ways to come up a ML model that predicts customers rating scores for songs Supervised data is available about customer's preferences in train.csv