A comprehensive method for attribute space extension for Random Forest

Attribute space extension for decision forests often contribute to improving the ensemble accuracy. In this paper we suggest the use of a recent method for attribute space extension where the newly generated attributes that have high classification capacity are chosen for extension. In literature, it is shown that the inclusion of these new attributes in the attribute space increases the prediction capacity of the decision trees. Random Forest is a state-of-the-art popular forest building algorithm that generates quite diverse decision trees. To increase the ensemble accuracy of Random Forest we consider the inclusion of more attributes with high classification capacity and employ the attribute extension technique that guarantees inclusion of newly generated attributes with higher classification capacity. We conduct an elaborate experimentation on ten different data sets from the UCI Machine Learning Repository. The experimental results show that ensemble accuracy for Random Forest increases when it is supplied with the aforementioned attribute extension technique.