Understanding e-learners' behaviour using data mining techniques

The information from Higher Education Institutions (HEIs) is primarily relevant for decision maker and educators. This study tackles e-learners behaviour using machine learning, particularly association rules and classifiers. Learners are characterized by a set of behaviours and attitudes that determine their learning abilities and skills. Learning from data generated by online learners may have significant impacts, however, few studies cover this resource from machine learning perspectives. We examine different data mining techniques including Random Forests, Logistic Regressions and Bayesian Networks as classifiers used for predicting e-learners' classes (High, Medium and Low). The novelty of this study is that it explores and compares classifiers performance on the behaviour of online learners on four variables: raise hands, visiting IT resources, view announcement and discussion impact on e-learners. The results of this study indicate an 80% accuracy level obtained by Bayesian Networks; in contrast, the Random Forests have only 63% accuracy level and Logistic Regressions for 58%.