Automatic Detection of Students Learning Style in Learning Management System

Abstract

Learning style is one of the major factors of student performance in any learning environment. Determining the learning style of students enhances the performance of learning process. This paper proposes an approach to classify students learning style automatically based on their learning behavior. One of the best widely used classifier algorithm is decision tree which is proposed in this paper. The main concern in decision tree classifier is the construction of significant rules which are required for accurately identifying learning styles. Lack of significant rules would result in misclassification of learning style. Hence, the main focus of this paper is to construct most significant rules which would strengthen the existing decision tree classifier to precisely and accurately detect the learning style of students. The student behavior is obtained from the web log files and then mapped with three learning dimensions of standard Felder Silverman learning style model. Subsequently, by employing significant rules in decision tree classifier, the student behavior has been automatically classified with high accuracy. This approach was experimented on 100 students for the online course created in Moodle Learning Management System. The evaluation result is obtained using inference engine with forward reasoning searches of the rules until the correct learning style is determined. The result is then analyzed with a confusion matrix of actual class and predicted class which shows that processing dimension shows variance whereas perception and input dimension were detected correctly with an average accuracy of 87%.