Study on credit evaluation of electricity users based on random forest
Yandong Zhao, Xiao Ma · 2017
In view of the problem of the risk assessment of electric customers, to improve the accuracy of the evaluation results, a new method of credit evaluation based on improved stochastic forest is presented in this paper. Through the method of secondary training to increase the weight of the decision tree leaves node's accuracy, reduce the impact of poorly classified decision trees on random forest, to improves the accuracy of prediction. This article used the data collected by Beijing electric power company, then compare the random forest, support vector machines, decision tree algorithm, the results show that the method has high accuracy, is promising in terms of electricity customer's application and development advantage.