Predicting Student Behaviors and Performance in Online Learning Using Decision Tree

Abstract

A practical problem in online learning and an important research problem in education is knowing how to predict student behaviors and performance in online learning, and implement school early warning based on the forecast results. In this study, we predict student behaviors and performance in online learning using a decision tree that uses a popular data mining strategy to construct a learning intervention model of an adaptive learning system. The results show that the male students have a highest probability of performing poorly in academics and poor total learning duration. Further, the master students have the highest probability of poor learning time span. Results also indicate that male science students have the highest probability of reduced average length of staying per study session, and the art students have the highest probability of limited discussion participation. Finally, the female art students have the highest probability of taking poor notes, while the male students have the highest probability of reduced feedback times.