Data Mining for Improving Online Higher Education Amidst COVID-19 Pandemic: A Case Study in the Assessment of Engineering Students

Abstract

Instructional materials, internet accessibility, student involvement and communication have always been integral characteristics of e-learning. During the transition from face-to-face to COVID-19 new online learning environments, the lectures and laboratories at universities have taken place either synchronously (using platforms, like MS Teams) or asynchronously (using platforms, like Moodle). In this study, a case study of a Greek university on the online assessment of learners is presented. As a testbed of this research, MS Teams was employed and tested as being a Learning Management System for evaluating a single platform use in order to avoid disruption of the educational procedure with concurrent LMS operations during the pandemic. A statistical analysis including a correlation analysis and a reliability analysis has been used to mine and filter data from online questionnaires. 37 variables were found to have a significant impact on the testing of tasks’ assignment into a single platform that was used at the same time for synchronous lectures. The calculation of Cronbach’s Alpha coefficient indicated that 89% of the survey questions have been found to be internally consistent and reliable variables and sampling adequacy measure (Bartlett’s test) was determined to be good at 0.816. Two clusters of students have been differentiated based on the parameters of their diligence, communication abilities and level of knowledge embedding. A hierarchical cluster analysis has been performed extracting a dendrogram indicating 2 large clusters in the upper branch, three clusters in the lower branch and an ensuing lower branch containing five clusters.