Predicting Student Performance to Reduce Dropout Using J48 Decision Tree Algorithm

Abstract

Student dropout is a major problem that faces most of the Private University of Bangladesh. In Bangladesh, a huge number of student studies in private universities. But, almost one-third of students are dropped out in the first year of their study. Various reasons are identified behind this problem, including the medium of study, assessment process and lack of knowledge in semesterbased education. In this research, the researcher uses the Educational Data Mining technique to predict the students' final grade after the Midterm Examination. This early result prediction can help the students, teachers and the university authority to take necessary action to reduce students' dropout. A lot of Education Data Mining tools are available. This research proposes a classification model particularly a decision tree algorithm to predict the future grades of the students in Introduction to Computer, a first-semester course for undergraduate university students. Popular data mining software, WEKA is used for model construction and evaluation.