Comparison of SVM and KNN Algorithms in Predicting Academic Interest of Management Study Program Students at Muhammadiyah University of Makassar

Abstract

Abstract Determining academic specialization for university students is a crucial stage in higher education because it directly influences study success and competency development. However, the process is often conducted subjectively and is not fully based on academic data. This study aims to compare the performance of Support Vector Machine and K-Nearest Neighbors algorithms in predicting academic specialization of Management students at Universitas Muhammadiyah Makassar. The dataset consists of core course grades from cohorts 2018 to 2021 that were preprocessed and labeled into three concentrations: Human Resource Management, Marketing, and Finance. The research method involved building classification models using SVM and KNN, which were evaluated using accuracy, precision, recall, and F1-score with various parameter settings and train–test splits. The results show that SVM with a Radial Basis Function kernel and a test size of 0.1 achieved the best performance with an accuracy of 70.55 percent. Meanwhile, KNN with k equal to five, Euclidean distance, and a test size of 0.1 obtained an accuracy of 57.53 percent. These findings indicate that SVM provides more stable and accurate classification than KNN for academic specialization prediction. Therefore, SVM is considered more suitable as a machine learning based decision support model for academic specialization purposes effectively.Keyword: Support Vector Machine, K-Nearest Neighbors, Machine Learning