Demographic performance in educational data mining for medical student academic success

Educational Data Mining (EDM) is a powerful approach for discovering hidden patterns over educational data as well as forecasting students' academic performance. This paper presents a novel approach with respect to Machine Learning (ML) methods in predicting the final test marks of undergraduate students’s using midterm exam grades as primary data. Despite employing demographic variables for predicting students performance using EDM has been deemed extremely problematic with respect to fairness as well as is now being critically explored in the area, but are commonly employed in practice with no significant reasoning. The implementation and discussion are based primarily on the notion of how they contribute to excellent model performance. In this study, the processes that make them significant for prediction has been examined as well as the implications for ideas of equity with Logistic Regression (LR) method. These findings emphasize the importance of learning more regarding the causes and effects behind the data and thinking cautiously about demographic aspects in each unique situation, as well as conducting additional research to understand how demographic features impact educational achievement.