Perbandingan Metode K-Nearest Neighbors Dan Support Vector Machine Pada Klasifikasi Lagu Daerah Dengan Penambahan Ekstraksi Fitur Principal Component Analysis

Abstract

This study examines the application of the K-Nearest Neighbor (KNN) and Support Vector Machine (SVM) algorithms in music data classification based on audio features on the IRSD (Indonesian Regional Songs Dataset) dataset. The main objective of this study is to evaluate the effect of using the Principal Component Analysis (PCA) dimension reduction technique on the performance of both algorithms. The experimental results show that the use of PCA with KNN improves classification accuracy, especially for the number of PCA components between 20 and 30 components. In the SVM model, the Radial Basis Function (RBF) kernel provides the highest accuracy, reaching 79%. Although PCA can reduce dimensions and speed up execution time, its effect on SVM is not significant. This study concludes that KNN with and without PCA can provide good classification results, while SVM with the RBF kernel is more stable and accurate. These findings provide insight for further research in the use of classification algorithms for more complex music audio data.