一种基于随机森林的停电敏感模型

A power outage sensitive model based on the random forest algorithm

为深入全面地对不同用电群体用户进行分析,实现停电敏感用户的精准识别,制定针对性的风险防控策略,有效减少客户来电风险,本文提出一种基于随机森林的停电敏感模型,对客户停电的敏感程度进行划分,进而实现差异化地运营管理客户,为营销部、设备部、客服中心等部门提供有效数据支撑,助力电网营销管理。本文将随机森林模型引入停电敏感预测中,并将预测结果与停电工单结合输出停电敏感高风险、中风险、低风险用户。在此基础上,以浙江湖州市2016年1月1日至2018年12月31日的数据为例进行了实例验证。模型结果显示,随机森林的预测结果准确性为88%,模型覆盖率为76.5%,模型的AUC值为0.77,结果优于逻辑回归和神经网络模型,模型的优良性为电网客户服务风险提供有力的数据参考。

In order to conduct in-depth and comprehensive analysis of different power users,achieve accurate identification of power outage sensitive users,formulate targeted risk prevention and control strategies,and effectively reduce the risk of customer calls,this paper proposes a power outage sensitive model based on random forest.The sensitivity of power outages is divided to achieve differentiated operation and management of customers,provide effective data support for marketing department,equipment department,customer service center and other departments,and help grid marketing management.In this paper,the random forest model is introduced into the outage sensitivity prediction,and the prediction results are combined with the outage electric bill to output outage sensitive high risk,medium risk and low risk users.Based on this,an example validation was conducted with data from January 1,2016 to December 31,2018 in Huzhou City,Zhejiang Province.The model results show that the accuracy of the prediction results of random forest is 88%,the model coverage is 76.5%,the AUC value of the model is 0.77,the results are better than logistic regression and neural network models,and the model’s excellent performance provides a strong data reference for grid customer service risk.