Identifying At-Risk Online Learners by Psychological Variables Using Machine Learning Techniques Academic Article

abstract

• The purpose of this study was to investigate a predictive model of online learners learning outcomes through machine learning. To create a model, we observed students motivation, learning tendencies, online learning-motivated attention, and supportive learning behaviors along with final test scores. A total of 225 college students who were taking online courses participated. Longitudinal data were collected over three semesters (T1, T2, and T3). T3 was used as training data given that it contained the largest sample size across all three data waves. To analyze the data, two approaches were applied: (a) stepwise logistic regression and (b) random forest (RF). Results showed that RF used fewer items and predicted final grades more accurately in a small sample. Furthermore, it selected four items that might potentially be used to identify at-risk learners even before they enroll in an online course.

authors

• Baek, Eunkyeng
• Kwok, Oi-Man
• Mcintosh, William
• Sweany, Noelle

published proceedings

• ONLINE LEARNING

author list (cited authors)

• Chien, H., Kwok, O., Yeh, Y., Sweany, N. W., Baek, E., & McIntosh, W. A.

complete list of authors

• Chien, Hsiang-yu||Kwok, Oi-Man||Yeh, Yu-Chen||Sweany, Noelle Wall||Baek, Eunkyeng||McIntosh, William Alex

publisher

• The Online Learning Consortium  Publisher

published in

• Online Learning  Journal

keywords

• At-risk Online Learners
• Logistic Regression
• Machine Learning
• Online Learning
• Random Forest
• Stepwise Regression

Digital Object Identifier (DOI)

• 10.24059/olj.v24i4.2320

start page

• 131

end page

• 146

volume

• 24

issue

• 4