ABSTRACT
Spotify is a music streaming platform that continuously updates its features
with various music variations. Fans of the Bangtan Boys group, also known as BTS,
are very popular and often listen to their songs. However, research on classifying
BTS songs based on mood using valence is rarely done. Each song has emotional
energy reflected in its energy level and is closely related to human psychology. The
problem faced by Spotify is the lack of features to listen to songs based on mood.
Categorizing pop songs based on mood can help users choose BTS songs that match
their feelings at certain times. This study aims to group pop music data based on 4
categories of mood from Thayer's model using the k-means and random forest
algorithms. The research method used is SEMMA, which includes the stages of
sample, explore, modify, model, and assess. The attributes used include title, artist,
release date, compatibility level, energy, key, loudness, mode, speechiness,
acousticness, instrumentalness, liveness, valence, tempo, ID, and duration. From
these attributes, data clustering is performed using the k-means algorithm and
performance calculations using the Davies-Bouldin index in the RapidMiner
application. The research results in four moods: angry, sad, calm, and happy. The
most common mood encountered is all moods except happy. Evaluation is done using
a confusion matrix, resulting in an accuracy rate of 98.64%.
Keyword: Spotify; BTS; Mood; SEMMA; K-Mean Clustering; Random Forest;
Rapidminer; Davies Bouldin Index; Confusion Matrix