Neural Network-Based System to Predict Early MOOC Dropout

• Published in: International Journal of Engineering Pedagogy Vol. 12; no. 5; pp. 86 - 101
• Main Authors: Sraidi, Soukaina, Smaili, El Miloud, Azzouzi, Salma, Charaf, Moulay El Hassan
• Format: Journal Article
• Language: English
• Published: 01.01.2022
• ISSN: 2192-4880; 2192-4880
• DOI: 10.3991/ijep.v12i5.33779

In recent years, MOOC (Massively Open Online Courses) revolu-tion has transformed the landscape of distance learning. Based on the distribution of educational content, this type of education is expected to undergo the same revolution as all of the traditional sectors of content and service sales, such as music, video, and commerce, due to the emergence of new technologies. How-ever, the completion rate remains a key metric of MOOC success as the number of students registering for a MOOC usually decreases during the course. This rate can reach 2 to 10% at the end of the course. Therefore, predicting dropouts is an excellent way to identify students at risk and make timely decisions. In this study, a prediction model is developed using one of the most widely used methods, the recurrent neural network (RNN). As a result, our model can be considered as an optimal option in terms of accuracy and fit for predicting dropouts in MOOCs.