Code for Predicting Dropouts in MOOCs

Massive open online courses (MOOCs) have developed rapidly in recent years, and have attracted millions of online users. However, a central challenge is the extremely high dropout rate — recent reports show that the completion rate in MOOCs is below 5% (Onah, Sinclair, and Boyatt 2014; Kizilcec, Piech, and Schneider 2013; Seaton et al. 2014). What are the major factors that cause the users to drop out? Can we predict who will dropout in a MOOC course?

We have collected MOOC data from XuetangX.com, and constructed two different datasets.

The first dataset, KDDCUP, contains 39 Instructor-paced Mode (IPM) courses and their enrolled students. It was also used for KDDCUP 2015. The other dataset contains 698 IPM courses and 515 Self-paced Mode (SPM) courses. The dataset contains richer information, which can be used to test the robustness and generalization of the proposed method. This dataset is referred to as XuetangX. All datasets are publicly available at http://www.moocdata.cn.

CFIN is a Context-aware Feature Interaction Network (CFIN) model for predicting users’ dropout behavior in MOOCs. CFIN utilizes a context-smoothing technique to enhance convolutional neural network (CNN). Attention mechanisms are also used to combine user and course information into the modeling framework. The algorithm can reach an accuracy of over 90% for predicting dropouts in MOOCs.

The code can be downloaded here, https://github.com/wzfhaha/dropout_prediction 

Wenzheng Feng, Jie Tang, Tracy Xiao Liu, Shuhuai Zhang, and Jian Guan. Understanding Dropouts in MOOCs. In Proceedings of the 33rd AAAI Conference on Artificial Intelligence (AAAI'19). [PDF]