大数据驱动下全接触渠道的电力客户精准画像  

Construction of Accurate Portrait System of Power Customers through Full Contact Channels Driven by Big Data

作　　者：陈辉 李艳 林思远 CHEN Hui;LI Yan;LIN Siyuan(Shenzhen Power Supply Bureau Co.,Ltd.,Shenzhen 518001,China)

机构地区：[1]深圳供电局有限公司,广东深圳518001

出　　处：《云南师范大学学报（自然科学版）》2023年第2期34-38,共5页Journal of Yunnan Normal University:Natural Sciences Edition

基　　金：中国南方电网有限责任公司科技基金资助项目(SZKJXM20190331)。

摘　　要：基于大数据平台和微服务架构,建立电力客户精准画像技术架构,形成大数据驱动下客户画像系统底层标签数据集合,结合模糊C均值聚类算法、决策树及XGBoost算法建立的电力客户标签,生成电力客户精准画像,构建电力客户精准画像体系.实验结果表明,所提方法能够精准挖掘电力客户数据,聚类多种客户标签,有效降低标签分类汉明损失,减小电力客户标签分类误差,确保生成有效电力客户画像.Based on the big data platform and micro-service architecture,the technical architecture of power customer accurate portrait is established to form the bottom label data set of the customer portrait system driven by big data.By using fuzzy C-means clustering algorithm,decision tree and XGBoost algorithm to predict the power customer labels established,the accurate portrait of power customers is generated and the accurate portrait system of power customers is constructed.Experimental results show that the proposed method can accurately mine power customer data,cluster various customer labels,effectively reduce the Hamming loss of label classification,reduce the classification error of power customer labels,and ensure the generation of effective power customer portraits.

关 键 词：大数据驱动 全接触渠道 电力客户 精准画像 模糊C均值 聚类算法 决策树 XGBoost 

分 类 号：TM714[电气工程—电力系统及自动化]