Exploring the World of Music

INFO 2950 Project

This study was conducted to explore how different factors contribute to the creation of popular songs. Through utilizing data provided by the Million Song Data Set created by Echo Nest, we were able to see information with respect to song popularity, tempo, duration, artist popularity, familiarity and much more. Our testing came to fruition in the form of several regression models and hypothesis tests. We investigated the impact of a song’s tempo and duration, as well as the artist’s popularity and familiarity to see what type of role they play in creating a popular song. At first we came to the idea that songs within a tempo range of 100-140 bpm would be the most popular songs would be under. We continued to see what role length of a song played and how duration effected how popular a song would become. The last few models investigate the role that an artist’s own popularity and the level of familiarity the public had with that artist changed a song’s popularity. Our results found that contrary to what we thought, musical factors along the lines of song length and tempo have a significantly lower impact on a song’s popularity compared to the role that is played by social factors such as familiarity and popularity of the artist. The following pages document the process we followed and in reaching our conclusions, and the data and models that accompany them.