Analysis of Music Features and Song Popularity Trends on Spotify Using K-Means and CRISP-DM

Abstract

Spotify, known as one of the best music streaming platforms, has played an important role in changing how listeners access, enjoy and interact with music. With millions of songs and extensive user data, Spotify provides an opportunity to understand listener behavior and the factors that contribute to a song's success and popularity. This research aims to examine the relationship between music features and the popularity of songs on the Spotify music platform by analyzing SSE values, Euclidean distance values, and cluster center values on the dataset attributes loudness, danceability, and energy. The framework used in this research is CRISP-DM (Cross-Industry Standard Process for Data Mining). The K-Means clustering algorithm and the Weka data mining application are used to decipher the features that influence the success and popularity of songs on Spotify. The research results show that groups/clusters 1, 2, and 3 are groups/clusters with songs that have high, medium, and low loudness, danceability, and energy respectively. Popular songs on Spotify are currently increasingly focused on loudness, danceability, and energy with a prominent trend, namely songs with high loudness, danceability, and energy are becoming more popular, while songs with low loudness, danceability, and energy are becoming less popular.