The Four Seasons: Identification of seasonal effects in LMS usage data

Abstract

Learning Management Systems (LMSs) are widely used by organizations to provide and manage educational activities. Particularly in higher education, the application of LMS platforms is well documented and evaluated in the literature for at least one decade, whereby evaluation is often restricted to user-oriented analysis of the acceptance, usefulness and usability and rarely relies on real data-sets. Previous research revealed that the usage patterns of web users and mobile users highly depend on the time period within a semester. Therefore, this paper specifically addresses the question how to identify and compare seasonal effects on the basis of an anonymized data-set. After proposing an Educational Data Mining based method for analyzing log files of LMS platforms and elaborating related work, we report a case study in which we compare the usage behavior of four different seasons. It shows that not only the intensity of platform usage but also certain activities of LMS users are highly dependent on the season. Moreover, seasons can be characterized e.g. through rank/frequency plots of n-grams or principal components of the browsing sessions in the period of time. The paper provides evidence that the detection of seasonal effects can be used for improving the navigation structures and personalization of LMS systems.