Taking Informed Action on Student Activity in MOOCs - Using Clustering to Find Meaningful Student Subgroups
Ralf Teusner, Kai-Adrian Rollmann, and Jan Renz
This paper presents a novel approach to understand specific student behavior in MOOCs. Instructors in MOOCs currently perceive participants only as one homogeneous group. In order to improve learning outcomes, they encourage students to get active in the discussion forum and remind the participants of approaching deadlines. While these actions are most likely helpful, their actual impact is often not measured. Additionally, it is uncertain whether such generic approaches sometimes cause the opposite effect for some students, as they are bothered with information irrelevant to them. On the basis of fine granular events emitted by our learning platform openHPI, we derive metrics and enable teachers to employ clustering, in order to divide the vast field of participants into meaningful subgroups which can then be addressed individually. We contribute a visualization tool called Cluster Viewer, that allows to interactively explore student activity, to take informed action and to measure the resulting outcomes of the action with respect to the tracked metrics. The approach was evaluated within a live course via several instructor interviews and was perceived helpful overall.