Machine Learning-Based Classification of Turkish Music for Mood-Driven Selection

Abstract

Music holds a significant role in our daily lives, and its impact on emotions has been a focal point of research across various disciplines, including psychology, sociology, and statistics. Ongoing studies continue to explore this intriguing relationship. With advancing technology, the ability to choose from a diverse range of music has expanded. Recent trends highlight a growing preference for searching for music based on emotional attributes rather than individual preferences or genres. The act of selecting music based on emotional states is important on both a universal and cultural level. This study seeks to employ machine learning-based methods to classify four different music genres using a minimal set of features. The objective is to facilitate the process of choosing Turkish music according to one’s mood. The classification methods employed include Decision Tree, Random Forest (RF), Support Vector Machines (SVM), and k-Nearest Neighbor, coupled with the Mutual Information (MI) feature selection algorithm. Experimental results reveal that, with all features considered in the dataset, RF achieved the highest accuracy at 0.8098. However, when the MI algorithm was applied, SVM exhibited the best accuracy at 0.8068. Considering both memory consumption and accuracy, the RF method emerges as a favorable choice for selecting Turkish music based on emotional states. This research not only advances our understanding of the interaction between music and emotions but also provides practical insights for individuals who want to shape their music according to their emotional preferences.