Predicting Student Retention in Smart Learning Environments Using Machine Learning

The emergence of web-based learning platforms through the smart- university era has provided distance learning opportunities for students and working professionals worldwide. However, this strategy of learning faces many challenges, such as increasing rates of student dropout and difficulty in monitoring online courses. To address these challenges, we present a data-driven approach in the form of classification problem that investigates the behavior of individual students in the virtual learning environment. Additionally, we identify students who are likely to succeed or show academic struggles during the course cycle. In this work, the machine learning model was trained by using a public dataset that collected from 30,000 students across seven courses. Specifically, the random forest algorithm is selected for developing an effective model that predicts students who are likely to succeed, whereas the regression model is used for identifying key factors that affect academic performance, such as completing online assignments and interacting on forum posts. The experiment results show the importance of using the presented approach to address the challenges of distance learning in the smart-university era.