Will I Remain Popular? A Study Case on Spotify

Abstract

Online streaming platforms are now the most important form of music consumption. In this paper, we present a model for predicting if a popular song on Spotify will remain popular after a certain amount of time. Spotify is the second biggest global streaming service. If a song is popular on this plataform it will ensure a good financial return for the artist and his label. We approach the problem as a classification task and employ classificators built on past information from the plataform's Top 50 Global ranking. The Support Vector Machine with linear kernel classificator reached the best results. We also verify if acoustic information can provide useful features for this problem.We made a series of classication rounds, where the results of one round were used as input of posterior rounds. Our results show that rankings previous data alone is sufficient to predict if a song will remain at the Top 50 Global two months in advance, achieving accuracy, negative predictive value, recall, specificity and F1 Score higher than 70\% for this task.

A Model for Predicting Music Popularity on Streaming Platforms

“…As an example of this competitiveness, Wikipedia catalogs 1,396 music labels only in the United States 3 . Note that, according to the Data Usa website, there are 138,000 artists on American lands 4 . A prediction model could give artists and labels an edge over competitors, because they could focus more on songs that tend to earn a good yield.…”

“…A previous version of this work was presented at the 2019 edition of the Brazilian Symposium on Computer Music (SBCM) [3]. During this study some models were developed using different approaches, like ranking positions [4] and acoustic characteristics of songs [5]. On the SBCM paper we presented the results of a model that predicts if a song on Spotify's Viral 50 Global ranking will appears on Top 50 Global ranking and vice-versa.…”