MOOC student dropout: pattern and prevention

Massive Open Online Course (MOOC) is a completely new education method which appeared in 2012. MOOC is great for life-long learning and educational resource sharing. Until now, there are more than twenty MOOC platforms online. Some studies pointed out that students, however, can hardly follow a course till the end on MOOCs. For most courses, less than 13% of students could follow. Such a high dropout rate will restrict development of MOOCs in the future. This paper shows an observation of MOOC data. A statistic analysis is applied to students' behavioral data. The result is discussed in the perspective of course design on MOOC platform. A general system for predicting students' dropout is developed. With unsupervised learning, the system can fit on different on-going courses. A discussion of this system with Data Structures and Algorithms course is conducted. We also apply the system to Introduction to Computing course to test scalability. Additionally, based on the research above, some suggestions for instructing students on MOOC are given.