Intelligent Pedagogical Systems for Higher Education: A Machine Learning-Driven Personalization Approach

Abstract

A reinforcement-learning-based pedagogical agent then dynamically adapts instructional strategies and learning materials based on the predicted learner state. In comparison to static instructional models and traditional classifiers, experimental assessment on a university-level Learning Management System (LMS) dataset of420 undergraduate students demonstrates quantifiable improvements in prediction accuracy, student engagement, and course completion. To facilitate independent verification, reproducibility information, statistical validation procedures, and plans for external-dataset validation are included with the main findings. The results imply that learning efficacy, learner happiness, and pedagogical decision-making in higher education institutions may all be significantly enhanced by machine learning-driven customisation.