Analisis Prediksi Mood Genre Musik Pop Menggunakan Algoritma K-Means dan C4.5

Abstract

Spotify is a music streaming platform that has a variety of diverse features and is always updated in terms of the latest music. The features in spotify have an interesting thing for users to enjoy music more optimally both in listening to songs based on songs, most popular artists and genres. Research on classifying songs based on mood by using energy and valence in a song is often done, especially in western pop songs. In every thought music has emotional energy that radiates and is strongly related to human psychology. The problem with spotify is that there is no feature to listen to songs based on mood. If pop songs are categorized by mood, it will be easier for people to listen to pop songs and choose the appropriate one based on mood. In this study, pop music data will be grouped based on 4 categories of Thayer's mood models using the k-means and c4.5 algorithms. The purpose of this study is to analyze the mood prediction of the pop music genre using the k-means and c4.5 algorithms. The research methodology used is SEMMA, the stages in Semma are sample, explore, modify, model and assess. The attributes used are danceability, energy, tempo and valence. From these attributes, data clustering is made using the k-means algorithm using RapidMiner. Then visualized using Power BI. The results of the research from cluster data are grouped into moods consisting of angry, sad, cheerful and happy. The most abundant mood is in the cheerful mood. Then evaluate the assess using the calculation of the confusion matrix which produces an accuracy rate of 91.9%..