Improved Student Performance Prediction Using Stacking Ensemble Learning with Gradient Boosting, Random Forest, and XGBoost

Accurately predicting student academic performance is crucial for early intervention and personalized education. This study presents an advanced predictive model employing a stacking ensemble framework that integrates Gradient Boosting, Random Forest, and XGBoost, with a Support Vector Regressor (SVR) as the meta-learner. The dataset includes an assortment of a categorical and numerical features including several demographic, and academic variables, with preprocessing (including OneHot and MinMax scaling) applied to all data to optimally represent features. The combined model was superior to individual algorithms by a large margin, with an accuracy-like metric (± 2) of 97.29%. (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>) score 0.751, and a MAE of 0.246. By leveraging the benefits of multiple ML models, this approach ensures that this technique is a large predictor of children in need of aid along with enabling data-driven teaching methodologies which is vital for both developing accurate and interpretable model in machine learning. The outcomes of a display what the employment of ensemble learning can better analytic student performance prediction and smart learning environments.