Comparison of Support Vector Machine and K-Nearest Neighbor Methods for Predicting Student Learning Outcomes Based on Student Performance Data

Abstract

Information systems and machine learning have encouraged the use of academic data to predict student learning outcomes in the education sector. This study evaluated the performance of Support Vector Machine (SVM) and K-Nearest Neighbor (KNN) algorithms in predicting student learning outcomes using the Student Performance in Exams Dataset from Kaggle. A quantitative computational experiment was conducted through several preprocessing stages, including target variable creation, categorical data conversion, feature standardization using StandardScaler, and data splitting into 80% training data and 20% testing data. Model performance was evaluated using accuracy, precision, recall, F1-score, Receiver Operating Characteristic (ROC) curve, and Area Under the Curve (AUC). The results show that SVM achieved an accuracy of 97%, with average precision, recall, and F1-score values of 0.97, while KNN achieved an accuracy of 90% and an F1-score of 0.88. The AUC values of 1.00 for SVM and 0.97 for KNN indicate that both models performed well, although SVM provided better class separation. Therefore, SVM was more effective than KNN in predicting student learning outcomes on the dataset used and may support the development of machine learning-based academic prediction systems.