Educational Data Mining and AI for Academic Performance Forecasting

Abstract

EDM and AI are emerging tools for assessing and forecasting academic achievement. This essay predicts academic success using AI utilizing machine learning and ensemble learning. This academic dataset covers socioeconomic and demographic data including semester grades, admission grades, scholarship or tuition fee status, parental qualifications, GDP, and inflation rate. SMOTE, encoding, normalizing, and feature selection minimize class imbalance and improve data quality. A recommended approach compares Decision Tree, Random Forest, Gradient Boosting, and XGBoost with 80:20, 70:30, and 50:50 train-test splits. Performance is measured by accuracy, precision, recall, and F1-score. Ensemble and boosting strategies surpass classification systems in experiments. XGBoost was most accurate at 94.10%, followed by Gradient Boosting at 91.20%.The novality of this research is to introduces Educational Data Mining and advanced ensemble learning methods including XGBoost, Gradient Boosting, and Random Forest to predict academic success. Preprocessing, SMOTE balance, feature selection, and dataset separation strategies improve prediction reliability in the proposed model. Demographic, socioeconomic, and academic aspects are also considered to construct a more robust and intelligent model for early student success prediction.