Using Academic Analytics to Predict Dropout Risk in E-Learning Courses

Abstract

Information technology is reshaping higher education globally and analytics can help provide insights into complex issues in higher education, such as student recruitment, enrollment, retention, student learning, and graduation. Student retention, in particular, is a major issue in higher education, since it has an impact on students, institutions, and society. With the rapid growth in online enrollment, coupled with a higher dropout rate, more students are at risk of dropping out of online courses. Early identification of students who are at risk to drop out is imperative for preventing student dropout. This study develops a model to predict real-time dropout risk for each student while an online course is being taught. The model developed in this research utilizes a combination of variables from the Student Information Systems (SIS) and Course Management System (CMS). SIS data consists of ten independent variables, which provide a baseline risk score for each student at the beginning of the course. CMS data consists of seven independent variables that provide a dynamic risk score as the course progresses. Furthermore, the study provides an evaluation of various data mining techniques for their predictive accuracy and performance to build the predictive model and risk scores. Based on predictive model, the study presents a recommender system framework, to generate alerts and recommendations for students, instructors, and staff to facilitate early and effective intervention. The study results show that the boosted C5.0 decision tree model achieves 90.97 % overall predictive accuracy in predicting student dropout in online courses.