An application with meta-methods (MetaRF) based on random forest classifier

Abstract

Meta classifiers are an area of intense study in the field of machine learning to improve classification performance. On the other hand, Random Forest is an important classifier in terms of providing fast and effective results. In this study, a meta-ensemble classifier called MetaRF based on the Random Forest basic learner is presented to use and combine the advantages of meta classifiers. For experimental results, the Random Forest base learner and eight meta-learners (AdaBoost, MultiBoostAB, Bagging, Stacking, UltraBoost, FeatureselectedClassifier, RandomSubSpace, FilteredClassifier) were used for ensemble classification on five datasets from the UCI Machine Learning Repository. Experimental results are promising in terms of accuracy rates, precision, recall and F-measure values. The method designed in the study is recommended to be used in machine learning studies and meta-classifier applications.