Educational Data Mining Based on Multi-objective Weighted Voting Ensemble Classifier

In recent years, data mining techniques have been widely applied in different topics such as education, medical, security, etc. In this research and in the first step, we apply educational data mining two well-known methods including J48 and Multilayer Perceptron (MLP). In addition, we will create some simple rules using J48. Our results in this part of research indicate that we get the best performance when class is “High”. We also show that most generated rules using J48 algorithm are related to “Low” class. By using generated rules, cut-off points presented that we observe that “STR” feature is not involved in the production of rules. Finally, by utilizing voting approach, we introduce a new Multi-objective Weighted Voting Ensemble Classifier (MWVE) using J48 as well as MLP algorithms which is called “MWVE-J48MLP” algorithm. According to our outcomes in this step, we observe that the proposed model has better performance than both J48 as well as MLP algorithms in previous step.