A clustered oversampling coupled random forest classification method

Abstract

To address the limitations of current classification prediction models, an algorithm DPC-ASMOTE-IRF for Clustering by fast search and find of density peaks (DPC) with adaptive SMOTE (ASMOTE) coupled with improved random forest (IRF) is proposed. The algorithm first introduces in the data processing stage DPC clustering algorithm to cluster the samples, and proposes ASMOTE oversampling technique to process a minority class of samples to get relatively balanced samples. In the algorithm stage, the model’s overall classification performance for unbalanced data is improved by assigning weights to each decision tree in the random forest through the classification accuracy. The experimental results show that the classification performance indexes of the proposed method are all improved compared with similar classification experimental models.