A prediction analysis on the dropout of cyber university based on learning analytics

Based on the learning analysis approach, this study was intended to predict the dropout of freshmen in the first and second semesters of cyber universities, and to compare the dropout patterns of freshmen in the first and second semesters. In order to predict the dropout, the data sets that can be collected in the course of learning were analyzed to distinguish the risk group of dropouts and suggested implications for preventing dropouts. Through the review of the preceding study, six factors were derived in three areas: learning conditions (economic burden, academic intensity), learning processes (learning participation activities, learning regularity), and learning outcomes (grade, class satisfaction) to derive a cyber university dropout prediction model through logistic regression. The analysis shows that the most important factors in predicting a cyber university s dropout are grades(GPA), learning regularity, class satisfaction, and full scholarship are proven predictors. As a result, the need to differentiate between the first semester learning support strategy for new students and the next semester as a preintervention to prevent them from going out of the university. In addition, regular learning was derived as a key strategy to continue the study without dropping out.