Ensemble Based Feature Selection With Hybrid Model

Abstract

In this study, a new mathematical model established with an ensemble-based approach, is proposed and applied to a large-scale data set consisting of three classes, whose features were extracted, obtained from birthday tweets. In this model, bagging method, which is one of the data variation methods, was applied first, and then a hybrid model combining the two approaches was created by applying the function variation approach obtained by using more than one feature selection method together. The resulting hybrid ensemble was first classified with the multi-class Support Vector Machines (SVM) algorithm, and then pruned with the ensemble pruning approach we propose in this study. By comparing the prediction success of the proposed model with the studies in the literature, it is observed that higher estimation success is obtained in comparison to those studies.