International Science Community Association Research Journal of Educational Sciences 14 2 2026 08 1 An explainable Ensemble learning approach for Student Performance Prediction 5 10 EN Patil Swati 1 Department of Management Studies, VIVA College of Commerce Science and Arts, Virar, Palghar, Maharashtra, India P. Sasikala 1 Department of Computer Science, Lal Bahadur Shastri Government First Grade College, RT Nagar, Bengaluru – 560032, Karnataka, India > Research Paper 2026 6 4 2026 08 1 Predicting students' academic performance at an early stage enables educational institutions to provide timely academic guidance and improve learning outcomes. Although machine learning techniques have been widely adopted for this purpose, many existing prediction models provide limited information about the factors influencing their decisions. The lack of model transparency reduces educators' confidence in applying artificial intelligence for academic decision-making. This study develops an explainable ensemble learning framework by combining Random Forest, XGBoost, and LightGBM with SHapley Additive exPlanations (SHAP). The framework incorporates data preprocessing, ensemble classification, and feature-level interpretation to generate accurate and explainable predictions of student performance. A publicly available dataset containing 14,003 student records with 16 academic, learning, and demographic attributes was used for model development and evaluation. The dataset was divided into training and testing subsets using an 80:20 ratio. The developed framework was evaluated using Accuracy, Precision, Recall, F1-Score, and Receiver Operating Characteristic–Area Under the Curve (ROC-AUC). The experimental findings indicate that the ensemble model achieves superior predictive performance compared with individual machine learning algorithms. SHAP analysis further identifies examination score, attendance, study hours, assignment completion, and motivation level as the most influential factors affecting academic performance. The proposed approach provides interpretable prediction outcomes that can support educators in recognising academically at-risk students, planning appropriate interventions, and improving educational decision-making. Copyright@ International Science Community Association