Term Project – Team 3 Topic – Spotify Top Tracks Katrina Sutton, Rilov Kulankara, Sean Lin
The data we have chosen to use is a list of Spotify’s most downloaded songs in 2018. It includes the name of the artist, song title as well as multiple song attributes created by Spotify to understand different variables about the songs. These song attributes are referred to as audio features and were extracted from the Spotify Web API.
The first objective we had for this assignment was to see if we could determine what in particular made a song popular. We wanted to know if there were any audio features that drove song popularity, then extended it further to see what those similarities were. Once we observed the similarities between attributes, we felt this would give us an understanding of why people liked these songs, and what people wanted to listen too. Lastly, we wanted to use our findings to see if we could use predictive modelling techniques to see if we could calculate whether or not a song would be popular based on these attributes.
The 2018 Spotify dataset we are using was derived from Kaggle. As part of the dataset there was categorical values such as the artist name and track title, but there were also numerous audio features that represented a numerical ranked value. These audio features were defined from Spotify; listed and defined below: