Student Participation Classification Using Support Vector Machine

Abstract

This study aims to classify student participation levels using the Support Vector Machine (SVM) algorithm. The data consists of discussion scores and the number of questions, with the target being student engagement status. The model was trained using three types of kernels (linear, RBF, and polynomial), followed by hyperparameter tuning using GridSearchCV. Performance evaluation was conducted using accuracy, precision, recall, and F1-score metrics, supported by decision boundary and confusion matrix visualizations. Results showed that the RBF kernel yielded the best performance, with 96% accuracy, 97% precision, 96% recall, and 98% F1-score. The application of SMOTE to balance the dataset further enhanced model performance. These findings confirm that combining SVM, parameter tuning, and data balancing techniques can produce an accurate and fair classification system. This model has potential to be integrated into decision support systems in education to enable early identification of disengaged students.