Interpretable Machine Learning for student performance prediction in “STEM” subjects in Ugandan Secondary Schools

Abstract

Predicting student performance in ”STEM” classes is a crucial endeavor that can aid teachers in mak ing better choices and supporting students more successfully. In this study, we provide a machine learn ing method for estimating secondary school students’ performance in STEM (science, technology, engi neering, and mathematics) courses. Our strategy is interpretable, which means it is made to be simple enough for human users—like teachers and administrators—to understand. In the context of educa tion, where openness and accountability are crucial, we first talk about the necessity for interpretable models. The dataset of student performance data that we use for our trials is then described. It was gathered from a selection of secondary schools in Uganda. Next, we present our interpretable machine learning model, which is based on a decision tree algo rithm. We explain the features and parameters of our model and how we trained it on the dataset. We also describe how we evaluated the performance of our model using standard metrics such as accuracy, precision, and recall. Finally, we present the results of our experiments, which demonstrate the effectiveness of our inter pretable machine learning model for predicting student performance in STEM subjects. We also pro vide some insights and recommendations for educators and policymakers in Uganda based on the in sights gained from our model. Overall, our work highlights the potential of interpretable machine learn ing to support student achievement and informed decision making in the education sector