Comparative Analysis of Student Performance Prediction Using Machine Learning Techniques

Predicting student performance is a critical task for educational institutions to facilitate timely interventions and improve academic outcomes. In this paper, we conduct a comprehensive comparative analysis of machine learning models applied to student performance prediction. The models evaluated include traditional approaches such as logistic regression, decision trees, and advanced techniques like XGBoost, Random Forest, and neural networks with attention mechanisms. Each model is assessed based on accuracy, precision, and F1-score. Dimensionality reduction techniques like t-SNE, along with feature selection methods such as feature importance, are employed to optimize the performance of the models. Our analysis shows that ensemble methods, particularly XGBoost, outperform other models in terms of accuracy, while attention- based models provide higher precision. We discuss the implications of these findings and suggest potential directions for future research, including the integration of real-time data and privacy-preserving techniques.