Using data mining techniques on Moodle data for classification of student?s learning styles

Building an adaptive e-learning system based on learning styles is a very challenging task. Two approaches to determine students learning style are mainly used: using questionnaires or data mining techniques on LMS log data. In order to build an adaptive Moodle LMS based on learning styles we aim to construct and use a mixed approach. 63 students from two courses that attended the same subject ?User interface? completed the ILS (Index of Learning Styles) questionnaire based on Felder-Silverman model. This learning style model is used to assess preferences on four dimensions (active/reflective, sensing/intuitive, visual/verbal, and sequential/global). Moodle keeps detailed logs of all activities that students perform which can be used to predict the learning style for each dimension. In this paper we have analyzed student?s log data from Moodle LMS using data mining techniques for classifying their learning styles focusing on one dimension of Felder-Silverman learning style: visual/verbal. Several classification algorithms provided by WEKA as J48 Decision Tree classifier, Naive Bayes and Part are compared. A 10-fold cross validation was used to evaluate the selected classifiers. The experiments showed that the Naive Bayes reached the best result at 71.18% accuracy.