Perbandingan K-Means dan GMM Untuk Analisis Popularitas Lagu Spotify Indonesia

Abstrak

The development of the music industry in Indonesia, especially through streaming platforms such as Spotify, creates challenges for record labels and musicians in making decisions regarding release timing and song promotion strategies. This study aims to compare two clustering algorithms, namely K-Means and Gaussian Mixture Model (GMM), in analyzing song popularity on Spotify Indonesia based on data such as total streams, peak streams, number of daily charts, and number of daily Top 10s. The results show that K-Means produces more accurate and easily interpretable clusters with a Silhouette Score of 0.731, while GMM has a Silhouette Score of 0.256, indicating less than optimal cluster separation. These findings indicate that K-Means is more suitable for data-driven decision making in the music industry, particularly for determining song release timing and promotional strategies. This research is useful for record labels, artist managers, and streaming platforms in designing more accurate and data-driven decisions.