Using an Ensemble Classifier on Learning Evaluation for E-learning System

Learning evaluation is an important part of cyber learning. Research studies aiming to increase the accuracy of performance evaluation in E-learning employ data mining technique. Here we describe a scheme for integrating classification algorithms that have been created by a machine learning method, trained on the real data set. The ensemble classifiers combine these classifiers, decision trees, neural networks, naive Bayesian into a single module and uses majority voting to fusion the output label. In this paper we discuss the theory behind the ensemble architecture, and present its implementation and a set of experiments using a variety of data sets. According to 10-fold cross validation bagging increases significantly the performance of every one classifier. Our work shows how the ensemble classifier performs remarkably for a specific data set on the application of E-learning evaluation system.