A Case Study of Interpretable Counterfactual Explanations for the Task of Predicting Student Academic Performance

One of the main topics of interest in the areas of Educational Data Mining and Learning Analytics is the prediction of students’ performance. To this end, a plethora of machine learning methods have been successfully applied to a range of educational data. Most of these methods, in order to increase the quality of their results, produce complex learning models, trading of their transparency of their inner mechanism. However, it is of utmost importance to develop understandable and explainable learning models. Comprehensive explanations linked with educational activities are valuable and give actionable insights to improve the learning outcome. In this context, the present study aims to examine the effectiveness of explainable and interpretable machine learning methods in the context of education. To demonstrate their practical usefulness, it presents a case study on interpretable counterfactual explanations. The results demonstrate the efficiency of the method and describe concrete steps that could be taken forwards in order to alter a model’s prediction.