Modeling learners' behavioral patterns and profiles, predicting the academic performance and investigating the effects of a dashboard in open and distance learning

This research aims to determine the patterns of learning materials use, and profiles of learners utilizing learning management system (LMS) in a massive open and distance education system. This research also seeks to develop a prediction model for academic performance, and design, develop and evaluate a learning-analytics-based learner performance dashboard integrated with the LMS. Conducted within the Open Education System of Anadolu University, the study utilized data belonging to enrolled learners that utilize the Anadolum eKampüs Learning Management System. The study used various data resources including learner demographic data, academic performance grades, data from navigation behaviors on the LMS, and finally self-report questionnaire data. Association rules were used to map the patterns of learning materials use. The outcome of the rules included previous exam questions and chapter summaries. Cluster analysis was utilized to map learner profiles. k-Means algorithm revealed 5 clusters depending on end-of-term average grades and average materials access. Machine and deep learning algorithms were used to develop prediction models for academic performance. Learner letter grades were predicted using the gradient boosted tree algorithm with an accuracy of 72.16%. The most important variables in developing the prediction model were mid-term grades and the rate of exam taking. Experimental study was conducted to investigate the effect of learner performance dashboard on learning processes. The tests revealed significant differences between the control and the experimental group in session time count, session day count, learning materials access count and academic performance.