A Modified Random Forest Based on Kappa Measure and Binary Artificial Bee Colony Algorithm

Random forest (RF) is an ensemble classifier method, all decision trees participate in voting, some low-quality will reduce the accuracy of random forest. To improve forest, with larger degree diversity and higher classification are selected for voting. In this paper, RF based on Kappa measure improved binary artificial bee colony algorithm (IBABC) proposed. Firstly, used pre-pruning, from Then, crossover operator leaping applied ABC, ABC secondary pruning, better performance The proposed method (Kappa+IBABC) tested a quantity UCI datasets. Computational results demonstrate that Kappa+IBABC improves most datasets fewer trees. Wilcoxon signed-rank test to verify significant difference between other pruning methods. addition, Chinese haze pollution becoming more serious. This predict weather has achieved good results.