JISE


  [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13]


Journal of Information Science and Engineering, Vol. 34 No. 5, pp. 1223-1235


Predicting Student Performance in MOOCs Using Learning Activity Data


YU-CHEN CHIU1, HWAI-JUNG HSU1, JUNGPIN WU2 AND DON-LIN YANG1,+
1Department of Information Engineering and Computer Science
2Department of Statistics
Feng Chia University
Taichung, 407 Taiwan
E-mail: {M0407060; hjhsu; cwu; dlyang}@fcu.edu.tw


  Massive Open Online Courses (MOOCs) allow students to study anytime, anywhere via the internet. Unfortunately, low completion rates and the enormous number of students make it difficult for instructors to monitor student progress. Students who perform poorly are susceptible to giving up due to a lack of appropriate counseling. Increasing the number of teaching assistants may ease the situation; however, this can be prohibitively expensive. In this study, we sought to improve the completion rate of MOOCs by predicting student performance through the analysis of data related to learning behavior and intervening before a student gives up. Learning behavior was first collected from OpenEdu, a well-known MOOC platform in Taiwan. A statistical model was then used to predict the performance of students based on this data. The effectiveness of the proposed model was demonstrated using cross validation of data from actual courses. Our findings demonstrate the effectiveness of monitoring student behavior in answering questions, watching videos, and participating in discussions on forums with the aim of predicting student performance in MOOCs in order to improve completion rates.


Keywords: MOOCs, data mining, machine learning, linear regression, logistic regression, software model generation

  Retrieve PDF document (JISE_201805_07.pdf)