Optimasi Sistem Rekomendasi Musik Berbasis Naïve Bayes: Studi Kasus pada Pengguna Musik di Spotify

Abstract

This study aims to develop a music recommendation system based on the Naïve Bayes algorithm, using Spotify users in Indonesia as a case study. The dataset was obtained from a questionnaire involving 473 respondents, covering variables such as gender, age, and frequency of Spotify usage. Music genres were grouped into two categories: Majority Favorites (Pop, K-Pop, Jazz, R&B, Indie) and Minority Favorites (Hip Hop, Rock, Religious, Dangdut, EDM, Regional). The research process included data cleaning and transformation, splitting the dataset into 80% training and 20% testing, applying the Naïve Bayes algorithm, and evaluating the model using accuracy, precision, recall, and F1-score metrics. The experimental results showed that the model achieved an accuracy of 95.82%, with 100% precision for the Majority Favorites category and 85.51% for the Minority Favorites category, along with recall values of 94.44% and 100%, respectively. The average F1-score was in the “very good” category, indicating that the model can reliably predict music genre preferences. These findings suggest that the resulting recommendation system is suitable for implementation to help users discover music aligned with their demographic characteristics and listening behavior, while also contributing to the development of recommendation systems based on primary data.