Ensemble Model for Educational Data Mining based on Synthetic Minority Oversampling Technique

Educational Data Mining (EDM) is a growing field that applies data mining, statistical analysis, and machine learning techniques to analyze student-related data. Existing EDM approaches often rely on manual statistical methods, which are time-consuming and less adaptable to dynamic educational environments. This paper proposes a novel ensemble-based framework that integrates machine learning classifiers with statistical approaches for student performance classification to address these limitations. To improve predictive accuracy, the model combines multiple classifiers, including Decision Tree, Logistic Regression, Random Forest, Multilayer Perceptron, and K-Nearest Neighbor. Given the inherent class imbalance in educational data, the Synthetic Minority Oversampling Technique (SMOTE) balances the dataset and enhances classifier performance. The proposed model is evaluated using a real-world dataset comprising 6,807 student records collected from a technological college in India. Performance is assessed using eight evaluation metrics to identify the most effective configuration. Results demonstrate the model's capability to deliver accurate and fair classification, aiding data-driven educational decision-making.