Student performance and time-to-degree analysis by the study of course-taking patterns using J48 decision tree algorithm

Abstract

Higher educational institutions (HEIs) are striving to improve their student's performance. Classification of students to predict student's performance could be useful in identifying bad performers or late completers to improve at an early stage by forecasting their final outcome. In this study, three student's performance prediction models were investigated. J48 decision tree algorithm was used for the study. When the performance of these models was evaluated, it was found that the third prediction model provided better performance compared to the other models. Also, there was variation in the course-taking patterns of students and contextual attribute course difficulty depending on their completion time. This demonstrates the significance of students course-taking patterns on performance and time-to-degree. The most significant finding is that unlike easy course delivering high GPAs interestingly some very difficult courses too lead to optimum time and GPA which could be attributed to other hidden factors which need to be investigated.