A Data-Driven Data Mining Framework Predicting CGPA and Identify At-Risk Students

Predicting student academic performance is crucial in higher education, as early identification of at-risk students can improve retention rates and academic success. This study presents a data mining framework that integrates K-means clustering, logistic regression, and linear regression to predict CGPA outcomes, identify at-risk students, and group students into performance-based clusters for targeted interventions. By analyzing early academic indicators such as SGPA and CGPA progression, this study offers a proactive approach to intervention strategies. The dataset, sourced from a public academic repository, was processed using data mining methods and validated with statistical metrics. The results demonstrate that early academic performance strongly correlates with final CGPA, enabling institutions to implement tailored academic support. K-means clustering revealed clear performance groups, logistic regression accurately classified at-risk students, and linear regression effectively predicted final CGPA. This integrated framework provides educational institutions with actionable insights to improve intervention planning and enhance student outcomes.