musikalkemist/musiccollaborativefiltering

simple example but not very useful

Opened this issue · 0 comments

Even though this shows, how to build a recommender system, it's not very useful, since the underlying data is quite poor.
R-squared is only 0.06 (1 being the highest).
I built something similar with more relevant data but still only had an R-squared of 0.50. To give useful recommendations something more is needed.