Comparative Analysis of Machine Learning Algorithms for Predicting Student Academic Performance in Higher Education

In recent years, educational institutions have increasingly sought to leverage data-driven approaches to enhance student success and retention. While numerous machine learning algorithms have been created to forecast academic performance, their efficacy differs among datasets, domains, and institutional contexts. The optimal algorithm for particular settings in higher education remains uncertain. This study aims to compare multiple algorithms such as Random Forest, Support Vector Machine, Decision Tree, Logistic Regression, Naive Bayes, and K-Nearest Neighbors to identify the most effective model for predicting student academic performance in higher education institutions (HEIs). This paper also introduces the application of SMOTE, which focuses on addressing class imbalance. Evaluation of multiple models comparing their performance across metrics such as accuracy, precision, recall, and F1-Score was present in this paper. The results of the comparative analysis showed that Random Forest outperforms other algorithms, particularly when SMOTE is applied, with an accuracy of 97%, demonstrating its robustness and adaptability in handling imbalanced datasets. This paper underscores the capacity of machine learning to furnish educational stakeholders with novel instruments for predicting student outcomes, facilitating prompt interventions to enhance academic performance.