An Advanced Explainable and Interpretable ML-Based Framework for Educational Data Mining

Livieris, Ioannis E., et al. “An advanced explainable and interpretable ML-based framework for educational data mining.“ International Conference in Methodologies and intelligent Systems for Techhnology Enhanced Learning. Cham: Springer Nature Switzerland, 2023. 

Over the past couple of decades, there’s been a surge in using advanced machine learning techniques to tackle complex issues in education. However, one significant challenge remains: many of these AI systems generate predictions without providing clear explanations for how they arrived at those conclusions. This lack of transparency hinders our ability to understand and trust the insights these systems provide. 

In our research, we’ve developed a novel framework aimed at predicting students’ academic performance while ensuring the results are not only accurate and reliable but also interpretable by humans. Our framework leverages the NGBoost algorithm, known for its efficiency in prediction modeling, and integrates it with techniques like LIME and SHAP. These methods help to unpack the black box of machine learning models by offering both local and global explanations for their predictions. 

Through various case studies, we’ve demonstrated the effectiveness of our framework and showcased how it can offer valuable insights for educators and students alike. By making the predictions more understandable and transparent, our approach aims to enhance the educational experience for all involved parties.