Early Prediction of Student Academic Performance Using Machine Learning and Clustering Techniques

Student academic performance prediction plays a crucial role in improving educational quality and enabling timely academic interventions. With the increasing availability of educational data, machine learning techniques have emerged as effective tools for analyzing student behavior and predicting learning outcomes. However, most existing studies focus on predicting final academic results, which limits their usefulness for early intervention.This paper proposes a machine learning–based framework for the early prediction of student academic performance by integrating clustering and supervised learning techniques. The proposed approach includes data preprocessing, feature selection, student clustering, hyperparameter optimization, and classification using multiple machine learning algorithms. Clustering is employed to group students with similar learning behaviors, followed by predictive modeling to identify at-risk students at an early stage. Experimental results demonstrate that ensemble-based models outperform traditional classifiers in terms of accuracy, precision, recall, and F1-score. The findings highlight the effectiveness of combining clustering and classification for early identification of academically at-risk students, enabling proactive academic support.