EARLY IDENTIFICATION OF AT RISK STUDENTS USING MACHINE LEARNING FOR GRADUATION PREDICTION

Abstract

This research employs machine learning to detect at-risk students in order to enhance graduation projections and institutional decisions. The analysis of academic achievement, attendance, demographics, behavior, and engagement data predicts early withdrawal and delayed graduation. Decision trees, logistic regression, random forests, and support vector machines are used to assess the accuracy and prediction of the model in order to identify the optimal fast intervention strategy. In order to retain and graduate students, educational institutions should invest resources in specialized academic aid, counseling, and mentorship after an early assessment of their needs. Traditional student monitoring methods can be repurposed to provide data-driven initiatives that improve learning and student performance.