A Hybrid Machine Learning Framework for Student Performance Prediction Integrating Academic, Demographic, Behavioural, Technology Access, and Psychological Indicators in Higher Education

Abstract

This study presents an explainable hybrid Random Forest (RF) and K-Nearest Neighbor (KNN) model for predicting student performance in higher education. The model integrates academic, demographic, behavioral, psychological, and technological access attributes to enable holistic analysis. Using a dataset of 1,500 students, the hybrid RF–KNN model achieved 96.3% accuracy, outperforming individual algorithms such as Decision Tree (J48) and Naïve Bayes. Explainable AI (XAI) techniques, including SHAP and LIME, were employed to interpret the feature contributions and provide actionable insights for educators. Results demonstrate that non-academic factors—particularly psychological and technological access—significantly enhance predictive performance and interpretability. This work aligns with Outcome-Based Education (OBE) principles by supporting early identification of at-risk learners.