Performance of Data Mining Classifiers on Kolb�s Learning Style Inventory (KLSI)

Abstract

Objectives: This study is intended to examine the effectiveness of predicting the learning style of college students. To understand experiential learning, many have reiterated the need to be able to identify students’ learning styles. Kolb’s Learning Style Model is the most widely accepted learning style model and has received a substantial amount of empirical support. Kolb’s Learning Style Inventory (LSI), although one of the most widely utilized instruments to measure individual learning styles, possesses serious weaknesses. Methods/Statistical Analysis: The proposed work introduces the study of efficiency in Kolb’s learning style. Classification algorithms used in this research include J48, BayesNet, and Naive Bayes and Random forest classifier. The data was collected from 30 students in Department of Information Technology, Dr. N.G.P. Arts and Science College in the final semester of academic year 2018. The 10-fold Cross Validation was used to create and test the model and the data was analysed by the WEKA program. Questionnaires were distributed to participants in class and then collected by the researcher after participants had finished them. Findings: After all the questionnaires were collected, data was then input in the computer and then statistically analysed. The J48, BayesNet, Naive Bayes and Random forest classifiers are used to measure the classification accuracy. In order to measure the learner style results the metrics like precision, recall, accuracy and Kappa values are used in this work. Application/Improvements: The result shows that random forest algorithm has more capability in predicting learning style of the students. Understanding the learning style of students is the first step and it may give a fruitful result when it is embedded with learning environment. Keywords: Kolb’s Learning Style Inventory (LSI), J48, NB Tree and Naive Bayes