Identifying Student Learning Styles Using Support Vector Machine in Felder-Silverman Model

Abstract

This research investigates how different questionnaire structures influence the accuracy of learning style predictions among students. The study examines various ratios of core and secondary questions, focusing on their impact on learning style classification. Utilizing Principal Component Analysis (PCA) for dimensionality reduction and integrating the Felder-Silverman Learning Style Model (FSLSM) with the Support Vector Machine (SVM) algorithm, the research identifies key insights into the effectiveness of these techniques. Results show that a balanced ratio of core to secondary questions, combined with PCA and SVM, significantly improves prediction accuracy, with Test 4 achieving the highest accuracy rate of 89.26%. This approach reveals that an optimal mix and quantity of questions can substantially enhance learning style classification. The findings underscore the importance of a tailored, question-specific strategy in personalized education and demonstrate the practical benefits of advanced machine learning methods. This study contributes to the field by providing a refined methodology for improving prediction accuracy and offers a roadmap for implementing more responsive and individualized learning experiences in educational settings.