An Enhanced Random Forests Model Based On Ordered Trees

Classification and regression models have been widely applied in many areas,such as economics,biology and insurance,etc.,hence it is significant to improve the predicting accuracy of the models.Random forest,one of the most prevalent algorithms in statistics and machine learning,is famous for its outstanding performance and concise theory.In this article,we first build a larger forest than before and take advantage of the out-of-bag samples to measure the accuracy of each tree in the forest,and then aggregate those more accurate trees to raise the predicting ability of the model.Afterwards,we propose the new model enhanced random forests.We also investigate the performance of the new method over a series of data sets and compare it with the traditional random forests,SVM and GBM models.It is competitive with other methods,and exactly performs better than traditional random forests.Moreover,we spend lots of time discussing how to select the two tuning parameters in the new model,and give some empirical suggestions finally.