Learner Behavior Profiling in a Learning Management System

Learning Management Systems lack features profiling learners based on their behavior such as learning styles and cognitive traits. Only a few models cited in the literature used the hybrid methods that combine multiple learner behavior such as learning styles and cognitive traits. This study proposed a methodology for identifying learner profiles in a learning management system based on learning styles and cognitive traits. Electronic course contents developed as online self-study modules for teaching a preclinical course were hosted on a university learning management system. The students accessed the learning materials for a 15-week semester. Two online continuous assessment tests were administered to students on the learning management system. The access records were extracted and mapped to the descriptions of the Felder Silverman Learning Style Model and the Cognitive Traits Model. Finally, the four clustering algorithms: k-means, hierarchical, PAM, Diana were used to analyze the extracted data. The clusters generated were validated using the silhouette coefficient method. Overall results of validation analysis showed a positive silhouette coefficient for all clusters indicating that all the students had been assigned to the right clusters. Taken together, these findings demonstrated that students with similar content navigation behavior shared common learning styles and cognitive traits. The student clusters generated can guide an instructor to use appropriate teaching methods suitable for each group. Furthermore, appropriate learning materials can be tailored for each group. 

Keywords: learning style; cognitive style; learning management system; learner modeling; learner behavior; machine learning; clustering