Combining Supervised and Semi-Supervised Models to Enhance Personalized Education

Personalized education has become an essential approach to addressing diverse learning needs and improving educational outcomes. This paper proposes a novel framework that combines supervised and semi-supervised machine learning models to enhance personalized education. Specifically, Random Forest is utilized to classify students based on their performance, engagement, and behavioral data, while Graph Neural Networks (GNN) capture and analyze the relationships between students, courses, and instructors in an educational graph. By leveraging both labeled and unlabeled data, this hybrid approach improves the accuracy of student risk predictions and enables the generation of customized learning path recommendations. The proposed framework was evaluated on a real-world educational dataset, demonstrating significant improvements in prediction accuracy and learning personalization compared to traditional methods. These findings highlight the potential of integrating supervised and semi-supervised learning techniques to create a more inclusive and adaptive educational environment.