Predicting learning effect by learner’s behavior in MOOCs

Abstract

With the fast development of MOOCs in recent years, more and more people start to take MOOCs to perfect themselves. However, there exist high dropout rate and low passing rate of examination in many courses. So it is very important to predict learners’ learning effect exactly. For learners who predicted good learning effect, teachers can impose intervention to help these learners to stick to the end of courses, while for predicted bad learning effect, teachers can take measures to help these learners to study harder to improve their learning. In this paper, we first analyze learners’ learning behavior data to explore the differences among learners with different categories, then a cascade prediction model is proposed to predict whether a learner can earn certificate in a course. Experiments conducted on a real-world dataset illustrated the effectiveness of the proposed model.