Course Recommender System Using Hybrid Machine Learning for Higher Education

Higher education institutions face a persistent disconnect between identifying at-risk students and providing timely, personalized academic guidance. This paper introduces a unified deep learning framework designed to bridge this gap by integrating dropout prediction with a pathway to course recommendation. Our novel hybrid architecture synergistically combines Recurrent Neural Networks (RNNs) to model structured academic histories with a DistilBERT transformer to interpret unstructured textual data. Trained on a public dataset from the UCI Machine Learning Repository, our model achieves a high predictive accuracy (93.7% ROC-AUC) for student dropout. More importantly, we demonstrate how the architecture is explicitly designed to be extended into a risk-aware recommender system, using the learned embeddings to suggest courses that align with a student's profile while mitigating dropout risk. While the component models are established, their integration into a dual-purpose system for educational intervention represents a significant step forward. Our work provides a validated, scalable framework for transforming predictive analytics into actionable, personalized student support.