Building a content-based music recommendation system

Question

I am trying to build a recommendation system in Python that recommends songs based on a playlist.

What I have is two datasets:
1. One dataset consists of 350 songs from my playlist and 13 acoustic features for each one like timber, energy, key, tempo, etc. that I extracted using the Spotify API
2. The other dataset consists of 340 000 songs and their acoustic features (got it from here: https://components.one/datasets/billboard-200/)
I've gotten both datasets in the same format and ready to be worked with.

Now what I am trying to do in Python is use my first dataset to get 30-40 songs with similar acoustic features from the second dataset but I have no idea how to approach this.
Should I use some machine-learning models or do something entirely else?
I thought about comparing the songs from my first dataset to the songs on my second dataset and pulling the ones with a similarity score of let's say over 70% or something like that but I feel like there is probably a much better way of doing this.

Answer 1

If the features are identical, good start would be to use n-neighbors approach.

It would be something like that

from sklearn.neighbors import NearestNeighbors
all_songs_features = [[0, 0, 2], [1, 0, 0], [0, 0, 1], [100, 100, 100], [0, 0, 1.5]]
neigh = NearestNeighbors()
neigh.fit(all_songs_features)

my_song_features = [[0, 0, 1.3], [1.1, 0, 0]]
print(neigh.kneighbors(my_song_features, 2, return_distance=False))

These code returns the indexes of similar songs from the bigger dataset.

Note, if the features are with different range (i.e. one with values 0-1 and another 10000-1000000), then you need to scale them, for example with StandardScaler