A Revision of LMS Interaction Classifications for Learning Analytics

Abstract

The choice of variables of interest for descriptive and predictive learning analytics usually relies on the data available for researchers and the application of machine learning algorithms on those data, which results on lack of insight and on case-specific solutions. The theory-based classification of variables emerges as a solution to this problem. This study reviews student interaction data classifications for learning analytics in Moodle LMS and proposes a new comprehensive categorization based on different learning cycle models: the Learning Cycle Interaction Categories (LCIC). LCIC comprises six different categories (engagement, content, knowledge validation (application), knowledge creation (dialogue/sharing), track/review and learning process management), whose combination may describe the different activities occurring during a typical session of a student within an LMS. The study uses InDash, a Moodle log data categorization, visualization, and analysis application, to exemplify the analysis of Moodle log data from one course. The results suggest that the combination of LCIC and InDash could be valuable for learning analytics, but also that further analysis is required to validate the adequacy of LCIC.