Spotify RecSys challenge

I’m still pretty new to the material but I figured I would let people know about this challenge from Spotify. Given lots of playlist data and corresponding metadata, the goal is to build a recommender system to recommend the next songs in the playlist.

I will probably play around with this some after doing a bit more material on collaborative filtering but it would be interesting to see what people on the forum can do with this!


Interesting. I wonder if a RNN would be a good choice here, maybe some sort of hybrid with embeddings on the songs * the given playlist used to filter the RNN input.