Detection of E-Learners’ Learning Styles: An Automatic Approach using Decision Tree *Aijaz Ahmed Kalhoro, Samina Rajper, Ghulam Ali Mallah Department of Computer Science, Shah Abdul Latif University, Khairpur Sindh, Pakistan Abstract—With the success of Information and Computer Technology (ICT), E-learning, as an efficient and effective mode of education, has captured attention from researchers and practitioners. The personalized e-learning systems, i.e., Adaptive Education Hypermedia systems (AEHS) are used to provide learning material, teaching instructions as per the student’s requirements. Class room Learning styles theories are being incorporated to resolve the traditional “One size fits all” issue of e-learning system. This study proposes automatic detection of learning styles from web logs of the students using the Data Mining technique, Decision Tree Induction. Kolb’s learning style theory is incorporated to understand e-learners’ learning styles on web. Using the decision tree for classifying different students into classes, the precision values for all four learning styles were achieved , i.e., Accommodator 0.652, Assimilator 0.5, Converger 0.574 and Diverger 0.509. The proposed approach can be used to detect the learning styles automatically from the web logs of e-learners to satisfy their learning needs to enhance e-learners' learning. Keywords-E-learning; Learning styles; Decision Tree; Data Mining I. INTRODUCTION E-learning systems are the training systems where through Internet technology, education can be provided to learners at their door steps. Several advantages of E-Learning are discussed in various research studies, i.e., Access of learning material anytime and anywhere [1], Unlimited discussions on discussion board [2], Collaborative activities like Discussion Boards, chats etc. [3]. But at the same time the e-learners are reported as they leave courses without completion for many reasons. It is reported by researchers that e-learners suffer due to lack of teachers’ interaction and diversity of students are dealt with the uniform teaching strategies. Therefore, the success of students on e-learning depends upon their own ability to manage the time and resources and to learn the material. Learning styles are helpful to understand the learning needs of the learners. Many personalized web based education systems have used learning styles to understand the learning needs and preferences of the e-learners on learning management system (LMS) [4-7]. Data Mining (DM) as a research field is contemporary to e-learning issues [8]. Data mining techniques extract valuable knowledge from the system logs; data generated by the users. In case of e-learning systems, the patterns can be found easily from web logs generated by teachers and learners [9]. Classification, sequential patterns, clustering, association, association and prediction are widely used DM techniques in educational DM [10]. A decision tree is widely used as a predictive model and a classifier to analyze patterns. This study proposes an automatic method to detect the learning styles of the e-learners using Data Mining technique, Decision Tree Induction [11]. The Kolb’s learning styles Model (KLSM) [12] will be incorporated to classify and detect the learners on the e-learning system. II. BACKGROUND AND RELATED WORK A. Learning Styles Detection on Personalized Web-Based Education Systems Adaptive Hypermedia Educational Systems (AHES) and Intelligent Tutoring Systems (ITS) are the ways for providing personalized e-learning services to the e-learners. Content presentation and the navigation are addressed in AEHS. While in ITS, an agent assisted services are provided to the e-learners. Both approaches need to know about the e- learners’ learning preferences. For understanding the students’ learning needs, various cognitive, psychological, educational theories are incorporated. Felder Silverman’s learning style theory [13] is used in personalized web based education systems, i.e., [5, 14-18]. However, [19] used Kolb’s learning style theory [12] using Bayesian Networks. To understand about the students’ learning styles on e- learning systems different e-learning systems use the learning style theories. To detect the learning styles into the e-learning systems different research studies adopted different techniques, i.e. questionnaire by CS383 [20], IDEAL [21], INSPIRE [22] and automatic detection techniques used by MANIC [23], SAVER [5], [24], [5],[19] to determine the students’ learning styles on LMS. Mostly Felder Silverman’s learning style theory [13] is used in adaptive hypermedia education systems and tutoring systems for understanding the students’ learning needs in the context of learning. Table 1. is used to show past research studies using Felder Silverman’s learning style model [13] Kolb’s learning style theory [12] is used to understand the learning styles of students on an e-learning system due to the reason that some of the research studies [25] [26] have reported unsatisfactory results about the mapped e-learning systems. Therefore, in this study another learning style is International Journal of Computer Science and Information Security (IJCSIS), Vol. 14, No. 8, August 2016 420 https://sites.google.com/site/ijcsis/ ISSN 1947-5500