Transforming Student Data Into Timely Interventions

Predicting academic performance is vital in education, yet early identification of at-risk students and timely interventions remain challenging. The authors present a decision support system using a Bayesian network to dynamically estimate course failure risk and integrate predictive analytics with institutional knowledge management. The model, built with the Bayesian Information Criterion and domain expertise, handles incomplete data and enables continuous, near-real-time monitoring. Applied to data from 8,897 undergraduates at a Southeastern U.S. university, it achieves over 85% accuracy at semester start and over 90% by midterm. The system enables prompt, individualized interventions and enhances institutional knowledge flows, supporting informed decisions and organizational learning. This scalable framework strengthens student retention and academic outcomes, bridging predictive modeling and knowledge-based strategies in higher education.