Explainable learning style classification using LIME and ensemble models in the VARK framework

This study investigates the use of explainable artificial intelligence (XAI) in classifying student learning styles through machine learning. A synthetic dataset simulating quiz-based interaction data was generated based on the VARK model (Visual, Auditory, Read/Write, Kinesthetic) to represent diverse learning preferences. Two state-of-the-art ensemble classifiers—Random Forest and XGBoost—were trained to predict learning styles from quiz responses and evaluated using standard classification metrics. To interpret the predictions, we employed LIME (Local Interpretable Model-Agnostic Explanations), producing both visual and HTML-based explanations at the individual level. Experimental results demonstrate high predictive performance and highlight LIME's ability to provide transparent, feature-level insights into model decisions. This work underscores the potential of XAI to support interpretable and personalized learning analytics in educational environments.