A Comparative Analysis of Dropout Prediction in Massive Open Online Courses

Abstract

Massive open online courses (MOOCs) provide a valuable learning platform for global learners. They are extensively utilized by an increasing number of people from all over the world due to their remarkable features, including unlimited enrollment, the lack of location requirements, free access to a high number of courses, and structural similarity to traditional lectures. However, high dropout rates negatively affect their educational effectiveness. In this regard, as a trending research topic in recent years, the prediction of dropout rates in MOOCs has become a critical issue in terms of planning for the future and taking precautions. This study proposes a practical prediction approach for the student dropout problem of MOOCs. In this regard, an adaptive neuro-fuzzy inference system (ANFIS) is utilized for the prediction of dropout rates in MOOCs for the first time in this study. The proposed approach uses the capabilities of both neural networks and fuzzy inference systems; thus, it provides highly accurate predictions. The performance of the proposed ANFIS approach is benchmarked against various models developed based on several machine learning methods, including the decision tree, logistic regression, support vector machine, ensemble learning, and K-nearest neighbor methods. The results reveal that the proposed approach provides higher statistical accuracy than its benchmarks, meaning that the proposed approach can be used effectively for MOOCs.