@article{Ibrahim_2025, title={Weighted Fusion of Machine Learning Models for Enhanced Student Performance Prediction}, volume={4}, url={http://dx.doi.org/10.22161/ijtle.4.6.11}, DOI={10.22161/ijtle.4.6.11}, abstractNote={<jats:p>Accurately predicting student academic outcomes is essential for enabling early interventions, improving learning support systems, and enhancing decision-making in higher education. This study proposes a weighted ensemble framework that integrates six machine learning models; Random Forest, Gradient Boosting, Logistic Regression, Support Vector Machine, Neural Network, and K-Nearest Neighbors to predict final grades using the Portuguese Student Performance dataset which contains 649 records. The weighting scheme is derived from each model’s validation performance, resulting in a balanced distribution where no single model dominates. The Random Forest model achieved the highest standalone accuracy of 76.92%, contributing most strongly to the ensemble, while the ensemble achieved 72.31% accuracy overall. Feature importance analysis across three interpretable models revealed that previous academic performance accounts for nearly 70% of predictive power, making it the strongest determinant of final outcomes. Behavioral factors such as absences, study time, and weekday alcohol consumption were significant secondary predictors, while social and demographic attributes provided additional contextual signals. The proposed model serves as an effective early warning mechanism for identifying at-risk students and guiding targeted academic interventions. This work contributes to data-driven educational analytics by combining interpretability, robustness, and actionable insights for educators and administrators.</jats:p>}, number={6}, journal={International Journal of Teaching, Learning and Education}, publisher={AI Publications}, author={Ibrahim, Ashraf Osman}, year={2025}, pages={104–109} }