基于物业状态画像的预付费电力客户辨识

Identification of Prepaid Power Customers Based on Property Status Portrait

向短租型物业推广预付费业务是提升电费回收率的有效途径。为解决租赁型物业查询困难的问题,从用电数据挖掘角度出发,首次提出一种基于物业状态画像的短租型物业辨识方法。首先提取与物业状态相关的用电特征,其次通过划分用电特征区间形成离散化标签,最后提出多级变权重聚类法完成物业画像,从中分离出作为预付费业务对象的短租型物业。算例结果表明,所提方法能有效辨识短租型物业,查准率和查全率分别达到85.8%和92.9%,可为电力公司推行针对短租用户群体的预付费等特殊供电服务提供信息基础。

It is an effective way to promote the prepayment business to the short-term rental property.In order to solve the problem of difficult query of rental property,this paper first proposes a identification method for the short-term rental property based on property status portrait from the perspective of power data mining.Firstly,the electricity consumption characteristics related to the property status are extracted.Then the discrete label is formed by dividing the electricity consumption characteristic interval.Finally,the multilevel variable weight clustering method is proposed to complete the property portrait,from which the short-term rental property is separated as the prepayment business object.The example results show that the proposed method can effectively identify the short-term rental properties,and the precision ratio and recall ratio are 86.1%and 90.8%respectively,providing information basis for power company to implement special power supply services such as the prepayment for the short-term rental user group.