A Case Study on Predictive Modeling for University Student Success and Underachievement: Utilizing R

The declining school-age population and evolving societal landscape necessitate innovative strategies in higher education to maintain institutional competitiveness. As universities transition towards a student-centered paradigm, data-driven decision-making emerges as a crucial tool. This study proposes a predictive modeling approach, leveraging random forest algorithms, to identify students at risk of academic underachievement and predict student success. To address the inherent imbalance in academic performance data, the study explores the potential of the Synthetic Minority Over-sampling Technique (SMOTE) to enhance model performance. By integrating fragmented data sources within university systems, this research aims to provide actionable insights for targeted interventions and support services. Additionally, the study delves into the practical implications of deploying predictive modeling as a service within the university, enabling data-informed decision-making at various levels.