Data Mining Technique For Early Detection Of At-Risk Students

Abstract:

In contrast to the past, the choice of class selection in the senior secondary school category has been an issue of major concern to all stakeholders in the educational sector. Among countless possible factors, the obvious ones are the parents/guardian deciding for their wards without considering their academic performance and ability to cope in the said class. However, the major challenge is how the educational administrators will assist parents/guardian likewise the students in making an informed decision on the change of selected arm of senior secondary school class as early as possible thus achieving better academic performance. In this study, the way of developing a student academic performance prediction model for early detection of at-risk students was looked into. To construct the model, five machine learning algorithms with different patterns were looked into. The attribute used was previous scores in related subjects and present score in the present class. The results revealed that random forest achieved an accuracy of 98.2%, which shows the potential efficacy of random forest as a predictive model for detecting at-risk students.

Keywords - Data Analytics, Data Mining, Student Academic Performance, Machine Learning, At-risk Students.