Predictive Modeling of Academic Success Using Machine Learning: A Comparative Analysis with SMOTE-Based Class Balancing

Abstract

This study evaluates six supervised machine learning algorithms, K-Nearest Neighbors (KNN), Decision Tree (DT), Logistic Regression (LR), Support Vector Machines (SVM), Random Forest (RF), and Artificial Neural Networks (ANN) for predicting academic outcomes (Graduate, Dropout, Enrolled) using an imbalanced real-world dataset. SMOTE was applied to address class imbalance. Random Forest achieved the highest accuracy (84.19%) after balancing and majority voting. Findings highlight the role of socio-economic factors and the effectiveness of ensemble learning in educational decision-making. The proposed model supports early interventions, especially in low- and middle-income academic contexts.