An independent record label is aiming to improve their artists album sales. To do it, they intend to apply Machine Learning in order to sculpt their songs in a more comercial way. The first attempt to do it is using Spotify Popularity feature. This feature measures a song success based on how many listenings a song had. The label firmly believe that some features can be good predictors of a song success and they want to know what are them. Also the label wants to predict with maximum accuracy how well a song will perform on Spotify every time a new song is releasen. We will create a model based on 19k songs available on Spotify and its main features, such as energy, key, loudness and many others.

The requirements for this project were NumPy, pandas, matplotlib, Seaborn and Scikit Learn.

All the modelling can be found on spotify_popularity_predictor.ipynb.