XAI-based dropout prediction and personalized recommendations in online education

This Master Final Work proposes an explainable, action-oriented framework that integrates machine learning–based dropout prediction with post hoc Explainable Artificial Intelligence (XAI) techniques to generate an automated personalized educational recommendation model. Using data from the UK Open University Learning Analytics Dataset (OULAD), the work develops and compares multiple machine learning models for dropout risk estimation, after which XAI techniques are applied to produce both global and individual-level explanations. By automatically operationalizing XAI outputs into personalized student recommendations, the findings demonstrate that incorporating explainability substantially enhances the practical value of dropout analytics by shifting the focus from risk prediction alone toward transparent, pedagogically meaningful interpretation and decision support in online learning contexts.