基于大数据分析的支撑智能催费的客户分群方法研究  

Research on Customer Grouping Method Supporting Intelligent Reminder Based on Big Data Analysis

作　　者：郑岳 韩娟 杜丽洁 于丽梅 仝天 孙源 ZHENG Yue;HAN Juan;DU Lijie;TONG Tian;SUN Yuan(State Grid Shandong Power Marketing Service Center(Measurement Center),Jinan 250001,Shandong,China)

机构地区：[1]国网山东省电力公司营销服务中心(计量中心),山东济南250001

出　　处：《电力大数据》2022年第8期55-61,共7页Power Systems and Big Data

摘　　要：电费催费作为电费回收的重要环节,具有面向用户数量多、客户需求差异大等特点,但是目前电费催费面临着传统催收模式下工作效率低、催收成本高和服务质量无法保证等问题,电费催收工作无法精准发力。因此本文对RFM用户价值模型进行了拓展和改进,在模型中延伸了欠费次数、欠费金额、年总电费等多个指标,能够有效评估用户缴费的用电情况,缴费及时性等。同时,通过对用户缴费、欠费等相关数据进行深入分析,筛选出用户缴费的特征指标,并提出了新的DC K-means聚类模型,在K-means聚类算法中引入密度的概念,有效解决选取聚类中心对最终聚类效果造成极大不良影响的问题,实现用户群体的分类,使得供电企业能够对不同类别的用户群体开展针对性的智能催费方式,以此来有效地提升催费效率,降低催费成本。As an important part of electricity bill recovery,electricity bill collection has the characteristics of a large number of users and large differences in customer needs.However,at present,electricity bill collection is faced with problems such as low work efficiency,high collection cost,and unguaranteed service quality under the traditional collection mode.The collection work cannot be carried out accurately.Therefore,through in-depth analysis of user payment,arrears and other related data,this paper filters out the characteristic indicators of user payment,and realizes the classification of user groups through clustering identification algorithm,so that power supply companies can carry out targeted intelligence for different types of user groups.In this way,the efficiency of electricity bill collection can be effectively improved and the cost of bill collection can be reduced.

关 键 词：大数据 聚类算法 智能催费 客户分群 

分 类 号：TM933[电气工程—电力电子与电力传动]