Modelling User Behaviour from Log Information in an Online Learning Management System

Abstract

An online Learning Management System (LMS) is a web-based technology used to organise, implement, and evaluate a learning process. In this context, the use of automated techniques is required to cluster the activities of students, allowing for better identification of their behaviours from large data sets. The initial phase in developing such an automated approach is to collect log information. An online LMS is suitable for this purpose, as it contains a large user behaviour data set. In this study, activity and process information are extracted from such a data set, followed by analysis of user behaviours to obtain various insights, such as the correlation between the learning process and the academic performance of students. By discovering processes, checking performance, and analysing engagement from the log data obtained from an online LMS, user behaviours can be successfully identified. In addition, the results demonstrate that the behaviours of students slightly affect their grades; for example, students with strong engagement in the online LMS tend to achieve higher marks. To support the significance of this study, a proposed scheme is provided along with examples and results. The key goal of this study is to identify student behaviours. By comparing different algorithms that can be applied in the context of this study, the results indicate that the k-Means clustering algorithm and Naïve Bayes classification algorithm are suitable methods for this purpose.