MOOC Learning Outcome Prediction Using Machine Learning Approaches

• In Event: Data Science in Education

Abstract

This study aims to predict learners’ learning outcomes (final grade in the course) by introducing demographic variables and some important MOOC learning behavior variables. The use of machine learning techniques in online educational research is increasingly important. Therefore, this study introduces and compares five state-of-the-art machine learning models: boosted logistic regression, Stochastic Gradient Boosting, Random Forest, K Nearest Neighbor, and Neural Network. The results indicate that boosted logistic regression, Stochastic Gradient Boosting, Random Forest, and Neural Network models attained high accuracy in predicting MOOC learning outcomes in both training and validation processes. This study offers insights into direction and intervention that online instructors may focus on to improve the MOOC learning environment and increase the learning outcome.

Authors

• Xue Wen, Louisiana State University

• Xuan Wang, The University of Texas Rio Grande Valley