Music recommender web service (ML)

For building music recommender, these data has been gathered from spotify api, cleaned and saved into csv.

Here is the repo to this story.

Algorithm

The recommendation is more than art than science; so the main engine of the recommendation has been built with many try and error, until it became a dependable music recommend engine. The engine could be found here.

As it can be seen I use multiple factors of musics and the compute the similarity based on each item then sum all similarity results.

This KNN implemented algorithm will return K item similar to asked song, and in the part 2 article the road to deploying it will describe.