Multidimensional Context Clustering to Analyse Student Engagement in Online Learning Environment

Abstract

Educational data mining is the application of data mining technology in an educational environment to indicate and resolve various types of issues faced in education. COVID-19 pandemic in 2020 accelerated the shift to emergency remote learning and online learning, which has continued to grow due to its flexibility and cost-effectiveness. However, several challenges exist when adopting the online learning, and which include the access to technology, technology attitude, psychological questions, teacher contact, and quality of assessment. Thus, it becomes important to focus on student engagement as a determinant of success during online learning. Student engagement is a complex construct which is comprised of four aspects which are behavioural, cognitive, emotional and social. In this study, the K-Means clustering approach is chosen to categorise students into clusters according to their active participation in the online learning process. The experiment used produces a silhouette coefficient of 0.71 clustering the datasets into three clusters. Cluster 0 are the disengaged learners who were observed to be least active across all the dimensions, while cluster 1 is composed of passive learners. The cluster 2 comprises the most engaged students with the high level of time management. These results provide information regarding the distinct engagement profiles that may be helpful when the lecturers attempt at student’s interventions.