Unsupervised Machine Learning Method for the Analysing of Students’ Activities in E-Learning

Abstract

Abstract:- The advancement of machine learning algorithms has made it possible for applying some of the unsupervised algorithms in clustering the activities of online students to analyse their learning behaviours as well as the implication of their learning behaviour to their final results. In this paper, students’ e-learning activities were automatically grouped into three (3) clusters using K-means clustering algorithms. Students actions were obtained from log files of their activities in Moodle Learning Management System (LMS). The aim is to group the students based on their similarities in actions on the LMS. Learning behaviours of students in each cluster were analysed and a correlation between each learning behaviour on student success or failure in their academic performance (Final Results) was investigated. The three clusters were labelled as Cluster A, Cluster B and Cluster C. The analysis shows that students in Cluster A have outperformed their counterpart parts with an average score of 91.12% in their final results, this group have higher login in the selected activities of the LMS, Cluster B with less number of interactions than Cluster A have average score of 75.65%. Finally, students in Cluster C have the least number of interactions with the LMS have failed the course with an average score of 36.57%. The research shows that, students who post, read and respond in Forum activities perform higher than those who do not. The work discovered that Forum activity has significant factor on student’s course success, however, this activity has less weight compare to other activities such as Assignment and Course View. The research suggests that weight should be allocated to Forum activities to encourage students’ participations.