Comparison of SVM, KNN, and NB Classifier for Genre Music Classification based on Metadata

Abstract

Music recommendations are one of the important things, such as music streaming platforms. Classification of music genres is one of the important initial stages in the process of music recommendation based on genre. Many music classifications are proposed by extracting audio features that require a not light computing process. This research aims to analyze and test the performance of music genre classification based on metadata using three different classifiers, namely Support Vector Machine (SVM) with radial kernel base function (RBF), K Nearest Neighbors (K-NN), and Naïve Bayes (NB). The Spotify music dataset was chosen because it has complete metadata on each of its music. Based on the results of tests conducted by the SVM classifier has the best classification performance with 80% accuracy, then followed by KNN with 77.18% and NB with 76.08%. The accuracy results are relatively the same as music classification based on audio feature extraction, so the classification with the extraction of metadata features can continue to be developed if the metadata in the dataset is well managed.