Dropout Detection in MOOCs: An Exploratory Analysis

The presence of MOOCs has increased exponentially in the context of distance education. However, many students who enroll in this type of courses drop out before completion. Several circumstances may cause dropout at any stage of the course and, in any case, before getting the certificate. Some students are not able to follow the course, others fail in the exams and leave, others only aim at getting a general overview of the course contents, others take all the activities but the final test, because of being interested in gaining knowledge but not in obtaining the certificate, etc. For this reason, it is interesting to analyze and understand the behavior of each student while interacting with the course. In this direction, the goal of this work is to predict whether a student will abandon a MOOC before completing it, so that it is possible to intervene accordingly, by warning the teacher about the dropout risk, notifying the student about this risk, etc. Different machine learning techniques have been tested with real data of a MOOC supported by EdX at UAM. In this article, the results of the work carried out in this direction are presented.