A COMPARATIVE STUDY OF VARIOUS MACHINE LEARNING ALGORITHMS IN PREDICTION OF DROPOUTS IN MOOCS

Abstract

The major challenge in Massive Open Online Courses is the huge difference between the number of enrolled learners and the registered learners. It has been observed through extensive literature survey that many registered learners do not complete their online course certification by not appearing in the proctored exam. A recent figures of NPTEL till Feb 2019 end showed that for 291 courses there was only 2.6% registration done. The huge number of dropouts of MOOCs could be predicted by incorporating various machine learning algorithms. This paper presents the comparison between various machine learning algorithms in terms of accuracy. The dataset used in this paper is collected as a part of the project completed for a SWAYAM online course titled Introduction to Learning Analytics offered in July 2019. Three machine learning algorithms named Decision tree, Nae Bayes and Logistic Regression have been compared in terms of their accuracy percentage. The dataset contains 4051 tuples and 16 attributes. The present study results could be used for building a recommender system or for performing prescriptive analytics for the predicted dropouts.