Support Vector Machine

The final cetegorical model I tried was an SVM model.

Creating Model

Github Python Notebook: SVM model


These are the features for my models:

  • Full data set: n=74,079
  • X =['danceability', 'energy', 'loudness', 'speechiness','acousticness','instrumentalness','liveness','valence','tempo']
  • y ='popularity' (might be popular, unpopular)
  • 30% test size
SVM model

This model proves to be somewhat accurate at predicting whether a song will be popular, but not to a useful degree. Its usefulness at predicting whether a song is unpopular is not useful either