Quantifying the likelihood of a song:
This notebook does basic analysis of song metadata taken from spotify. The data contains numeric metrics generated by spotify which measure the songs' danceability, mood, liveness, etc. The data also contains the songs' title and artist.
Loading the dataset
Looking for missing fields
Deciding factor variables
Plotting scatterplot matrix to understand the correlation
Plotting Decision Trees
Calculating the prediction accuracy of the decision trees
Generating Random Forests
Calculating the prediction accuracy of random forests
Random Forests perform better with a prediction accuracy of 80%.
According to random forect algorithm, instrumentallness and loudness are almost equally important variables to predict the "liking" of the song.