An Explainable AI Framework for Enhancing Pedagogical Decision-Making Through Learning Analytics

Abstract

Learning analytics, as well as Explainable Artificial Intelligence (XAI), have emerged as key approaches for improving educational decision-making by analyzing student interaction data and providing interpretable insights. The current study focuses on developing an explainable AI framework to enhance pedagogical decision-making through learning analytics using the EdNet-KT3/KT4 dataset. The proposed method integrates machine learning models with SHAP and LIME techniques to predict student performance and generate transparent explanations of learner behavior. The framework is assessed using a variety of performance indicators to guarantee reliability and interpretability. The main result demonstrates that the suggested model attains an accuracy of 96.12%, outperforming traditional models while maintaining full explainability. The findings demonstrate that the integration of XAI improves both prediction performance and pedagogical usability by enabling instructors to identify at-risk learners and support data-driven interventions. The study concludes that explainable learning analytics provides an effective and trustworthy approach for modern educational environments and adaptive learning systems.