基于回归算法的全量客户消费预警模型  

Early Warning Model of Total Customer Consumption Based on Regression Algorithm

作　　者：张靖 侯晓晶 ZHANG Jing;HOU Xiaojing(China Mobile Communications Group Shanxi Co.,Ltd.,Shuozhou Branch,Shuozhou 036002,China)

机构地区：[1]中国移动通信集团山西有限公司朔州分公司,山西朔州036002

出　　处：《数字通信世界》2021年第8期10-13,共4页Digital Communication World

摘　　要：近年来,移动手机用户使用流量大幅提升,但由于流量单价下降,语音需求放缓,导致通信行业收入增长乏力。为监测收入变化,预警用户的消费异动,及时采取维稳措施,基于回归分析算法,构建了全量客户消费预警模型,通过线性拟合、二次函数拟合,对个体用户的消费、流量、语音建立预测方程式,通过个体收入的预测实现对整体收入的预测,相对误差在5%以下,并根据用户消费变化趋势,输出重点客群,辅助市场营销,有效缩小了目标客户规模,平均营销成功率为12.36%。In recent years,mobile phone users have seen a signifi cant increase in traffi c usage,but due to the decline in the unit price of traffic and the slowdown in voice demand,the revenue growth of the communications industry has been weak.In order to monitor income changes,warn users of consumer changes,and take timely measures to maintain stability,this thesis builds a full-scale customer consumption warning model based on regression analysis algorithms,and uses linear fitting and quadratic function fi tting to assess individual users’consumption,traffi c and voice.Establish a forecasting equation,realize the forecast of the overall income through the forecast of individual income,and output key customer groups according to the changing trend of user consumption to assist marketing.

关 键 词：回归分析 消费预警 通信 

分 类 号：TN929.53[电子电信—通信与信息系统] TP3-05[电子电信—信息与通信工程]