Academic Performance Classification: Use of Supervised Learning Approach in Educational Data Mining

This study reviews the published studies that predict student performance or classify them into performance groups relying on different types of education-related data coming from diverse sources. This review study aims to determine the types of, sources of, and size of data used in educational data mining studies. It also aims to find out the distribution of supervised machine learning models, and tools/software used in educational data mining studies. In order to achieve these goals, 139 relevant publications (i.e., academic journal articles, and conference papers) are located in the Web of Science Citation Index database for review using some including/excluding criteria. Then, each paper is reviewed. The findings suggested that classification studies in educational data mining mostly relies on conventional machine learning algorithms using students’ education record and course/learning activity logs as predominant features for predicting students’ performance or classifying them into performance groups.