基于随机森林与RFM模型的财险客户分类管理研究
|
摘要
基于客户价值的财险客户分类管理能够帮助公司更有效地节约成本,创造收益。通过在RFM(近度、值度、频度)模型中加入财险客户理赔额指标,将模型扩展为RFMP模型,综合考虑了财险客户的利润贡献度及其风险因素,从风险和贡献两个角度更有效的衡量客户价值。同时,将随机森林分类算法应用到客户分类管理中,并与神经网络分类模型进行比较,实验结果显示随机森林分类具有更小的误差。进一步分析了各客户类群的人口统计学指标,避免了复杂的客户指标量化计算过程,有利于财险公司对庞大的客户群进行分类管理,也有助于公司对新入客户进行风险价值评估,提供具有针对性与个性化的产品与服务。
Classification management of insurance customers based on customer values can help companies to save costs and create revenue more effectively. We add index of claim amount of insurance customers to the RFM( recency,frequency,monetary) model to extend the model as RFMP model. The adjusted model considers both the profit contribution and risks of customers,and measures the value of customers more effectively from the two angles of risk and contribution. Meanwhile,Random Forest classification algorithm is applied to the customer classification management,and is compared with neural network classification model; the experimental results show that the Random Forest classification has smaller error. Demography index of various customer groups is further analyzed; complex customer index quantification and calculation process is avoided; it is beneficial for the insurance company to conduct classification management on the giant customer groups; it is also helpful for the risk value evaluation of the new customers to provide targeted and personalized product and services.
Classification management of insurance customers based on customer values can help companies to save costs and create revenue more effectively. We add index of claim amount of insurance customers to the RFM( recency,frequency,monetary) model to extend the model as RFMP model. The adjusted model considers both the profit contribution and risks of customers,and measures the value of customers more effectively from the two angles of risk and contribution. Meanwhile,Random Forest classification algorithm is applied to the customer classification management,and is compared with neural network classification model; the experimental results show that the Random Forest classification has smaller error. Demography index of various customer groups is further analyzed; complex customer index quantification and calculation process is avoided; it is beneficial for the insurance company to conduct classification management on the giant customer groups; it is also helpful for the risk value evaluation of the new customers to provide targeted and personalized product and services.
引文
[1]孙瑛,马宝龙,李金林.基于RFM模型方法的忠诚计划会员顾客价值识别研究[J].数学的实践与认识,2011(15):75-79.
[2]王文贤,金阳,陈道斌.基于RFM模型的个人客户忠诚度研究[J].金融论坛,2012(3):75-80.
[3]吕斌,张晋东.基于RFM模型的商业银行营销决策分析[J].统计与决策,2013(14):65-67.
[4]杨彬.一种基于RFM模型数据挖掘处理双阶段客户关联分类方法[J].统计与决策,2015(7):77-79.
[5]HAMID K.A New Application of RFM Clustering for Guild Segmentation to Mine the Pattern of Using Banks'e-Payment Services[J].Journal of Global Marketing,2014,27(3):178-190.
[6]DURSUNA A,CABER M.Using data mining techniques for profiling profitable hotel customers:An application of RFM analysis[J].Tourism Management Perspectives,2016(18):153-160.
[7]HOBLEY EU,BALDOCK J,WILSON B.Environmental and human influences on organic carbon fractions down the soil profile[J].Agriculture,Ecosystems and Environment,2016,223:152-166.
[8]BECKSCHFER P,FEHRMANN L,HARRISON R D.Mapping Leaf Area Index in subtropical upland ecosystems using Rapid Eye imagery and the random Forest algorithm[J].i Forst:Biogeosciences and Forestry,2014(7):1-11.
[9]GOUNARIDIS D,KOUKOULAS S.Urban land cover thematic disaggregation,employing datasets from multiple sources and Random Forests modeling[J].International Journal of Applied Earth Observation and Geoinformation,2016,51:1-10.
[10]李欣海.随机森林模型在分类与回归分析中的应用[J].应用昆虫学报,2013(4):1190-1197.
[11]明均仁,肖凯.基于R语言的面向需水预测的随机森林方法[J].统计与决策,2012(9):81-83.
[12]马玥,姜琦刚,孟治国,等.基于随机森林算法的农耕区土地利用分类研究[J].农业机械学报,2016(1):297-303.
[13]蔡加欣,冯国灿,汤鑫,等.基于局部轮廓和随机森林的人体行为识别[J].光学学报,2014(10):212-221.
[14]孙菲菲,曹卓,肖晓雷.基于随机森林的分类器在犯罪预测中的应用研究[J].情报杂志,2014(10):148-152.
[15]方匡南,朱建平,谢邦昌.基于随机森林方法的基金收益率方向预测与交易策略研究[J].经济经纬,2010(2):61-65.
[16]于晓虹,楼文高.基于随机森林的P2P网贷信用风险评价、预警与实证研究[J].金融理论与实践,2016(2):53-58.
[17]SINGH S,SINGH S.Accounting for risk in the traditional RFM approach[J].Management Research Review,2015,39(2):112-123.
[2]王文贤,金阳,陈道斌.基于RFM模型的个人客户忠诚度研究[J].金融论坛,2012(3):75-80.
[3]吕斌,张晋东.基于RFM模型的商业银行营销决策分析[J].统计与决策,2013(14):65-67.
[4]杨彬.一种基于RFM模型数据挖掘处理双阶段客户关联分类方法[J].统计与决策,2015(7):77-79.
[5]HAMID K.A New Application of RFM Clustering for Guild Segmentation to Mine the Pattern of Using Banks'e-Payment Services[J].Journal of Global Marketing,2014,27(3):178-190.
[6]DURSUNA A,CABER M.Using data mining techniques for profiling profitable hotel customers:An application of RFM analysis[J].Tourism Management Perspectives,2016(18):153-160.
[7]HOBLEY EU,BALDOCK J,WILSON B.Environmental and human influences on organic carbon fractions down the soil profile[J].Agriculture,Ecosystems and Environment,2016,223:152-166.
[8]BECKSCHFER P,FEHRMANN L,HARRISON R D.Mapping Leaf Area Index in subtropical upland ecosystems using Rapid Eye imagery and the random Forest algorithm[J].i Forst:Biogeosciences and Forestry,2014(7):1-11.
[9]GOUNARIDIS D,KOUKOULAS S.Urban land cover thematic disaggregation,employing datasets from multiple sources and Random Forests modeling[J].International Journal of Applied Earth Observation and Geoinformation,2016,51:1-10.
[10]李欣海.随机森林模型在分类与回归分析中的应用[J].应用昆虫学报,2013(4):1190-1197.
[11]明均仁,肖凯.基于R语言的面向需水预测的随机森林方法[J].统计与决策,2012(9):81-83.
[12]马玥,姜琦刚,孟治国,等.基于随机森林算法的农耕区土地利用分类研究[J].农业机械学报,2016(1):297-303.
[13]蔡加欣,冯国灿,汤鑫,等.基于局部轮廓和随机森林的人体行为识别[J].光学学报,2014(10):212-221.
[14]孙菲菲,曹卓,肖晓雷.基于随机森林的分类器在犯罪预测中的应用研究[J].情报杂志,2014(10):148-152.
[15]方匡南,朱建平,谢邦昌.基于随机森林方法的基金收益率方向预测与交易策略研究[J].经济经纬,2010(2):61-65.
[16]于晓虹,楼文高.基于随机森林的P2P网贷信用风险评价、预警与实证研究[J].金融理论与实践,2016(2):53-58.
[17]SINGH S,SINGH S.Accounting for risk in the traditional RFM approach[J].Management Research Review,2015,39(2):112-123.