Virtual Booth

High attrition rates in challenging STEM disciplines, particularly in courses like Financial Mathematics, demand proactive educational solutions. This project introduces a novel Predictive Academic Early Warning System powered by Machine Learning and Association Rule Mining (ARM). Moving beyond traditional reactive academic advising, our solution analyzes the historical academic trajectories of students to uncover hidden dependencies between pre-university STEM preparation and advanced course success. By processing data from Diploma in Actuarial Science students, the system successfully extracts over 3,385 highly accurate predictive patterns. The core innovation of this product lies in its ability to identify at-risk students with a critical 2-3 semester lead time, allowing educators to deploy targeted interventions well before a student fails. Furthermore, the model provides interpretable, data-driven insights by demonstrating, for example, that students with early mathematical proficiency are 3.2 times more likely to excel. By combining the interpretability of ARM with the predictive power of machine learning, this framework offers educational institutions a scalable, practical tool to address knowledge gaps, optimize targeted support, and ultimately improve STEM student retention and success rates.