加油卡客户流失分析和预测——基于“跨行业数据挖掘标准流程”  

Analysis on fuel card customer churn based on CRISP-DM

作　　者：高昉 GAO Fang(Shanghai Urban Construction Vocational College)

机构地区：[1]上海城建职业学院

出　　处：《国际石油经济》2019年第10期99-105,共7页International Petroleum Economics

基　　金：教育部人文社会科学研究一般项目（16YJE630002）;上海城建职业学院教育科学研究项目（cjyb201901）

摘　　要：采用跨行业数据挖掘标准流程(CRISP-DM)进行加油卡客户流失影响因素分析和流失倾向预测,建立了一个预测客户流失的决策树模型。该模型具有非常好的准确率、命中率和查全率,可帮助成品油销售企业及时发现有可能流失的加油卡客户,有针对性地开展客户维系与挽留工作,最大程度地减少加油卡客户流失。Based on CRISP-DM(cross-industry standard data mining process), the paper analyzes influencing factors of customer loss of refueling card, predicts the tendency of customer loss, and establishes a decision tree model with good forecasting results. With the high accuracy, hit rate and recall rate, the model will help refined oil product sales enterprises to timely find customers who may lose fuel cards, carry out targeted maintenance and retention of customers, and minimize the loss of fuel cards customers.

关 键 词：客户流失 CRISP-DM 加油卡 数据挖掘 决策树 

分 类 号：F27[经济管理—企业管理]