基于数据挖掘技术的人寿保险品质管理系统
|
摘要
保险行业在中国是一个快速发展的朝阳行业,业务的急剧增长带来了更多的客户信息。随着保险公司的网络速度和服务器处理能力不断提高,基于数据的商业智能业务分析也逐渐成为趋势。近几年保险业在中国快速发展,客户量不断上升,保险公司的各项运营成本也在大幅度增加。保险公司的业务收入一方面是来自新客户的保费,更重要的是续期客户的保费收入。续期保费收入具有比较高的稳定性和连续性的特点,是保险公司利润的主要来源,是保险公司持续发展的坚实基础。目前保险公司续期收费的方式比较单一,对客户采取同样的催交和通知方式,管理方式粗放,不利于保险风险和成本控制。
本文基于数据挖掘分类方法,分析现有客户的缴费情况,发掘交费概率和交费次数以及交费准时程度的关系,将客户细分为优质客户、准优质客户、忠诚客户、非稳定客户等几类。利用决策树和多元线性回归方法,建立续期客户交率的预测模型,指导续期部门对不同客户采取差异化的催交策略。使保险公司客户管理更有针对性,降低客户保费迟缴、忘缴等风险,并降低管理成本。同时,利用聚类方法,从大量的客户中分析具有较高退保风险的客户,采取相应措施,尽量降低退保事件发生的概率,从而提高续期收费的成功率,提高公司的保单品质,确保公司长期经营发展。
在充分考虑保险公司实际经营需求基础上,将本文所研究的客户续期保单分析模型进行实现,并设计与实现人寿保险品质管理系统。该系统在太平人寿四川分公司顺利应用,系统运行效果好。为该公司保单续期工作的高效开展、控制退保风险和运营成本、提升客户服务和业务品质提供了有力的工具和准确的数据参考。
The insurance industry in China is a rapidly developing industry, business grows quickly to bring more customer information. Along with the insurance company's network speed and server processing capacity continue to improve, based on the data of business intelligence and analysis has gradually become a trend. In recent years, insurance industry in China develops rapidly, the rising volume of customers, the insurance company of the operating costs have increased by a large margin. Insurance company business income is from first-year premium, more important is the renewal premium. Renewal premium has higher stability and continuity characteristics, is the main source of profit, and is a solid foundation for insurance company. At present the way of getting renew premium in insurance company’s very simple,for every customer,they give notice in the same way,it cost a lot and waste a lot. does not favor the insurance risk and cost control.
This thesis try to use data mining technique and classification method to analyze the activities of customer in payment,to discover the relationship between the possibility of payment and how many times they have paid and when they have paid the premium. According to the information of customer,we partition the customer in several group,such as high quality customer,quasi high quality customer,faithful customer and risk customer.Using decision tree and multi-variable regression, we make a prediction model to estimate the possibility of a certain customer’s activity in payment . guide the renewal sectors for different clients to take the difference of expediting strategy. The insurance company customer management more targeted, reduce customer premium late payments, forgot to pay, and reduce the cost of management. At the same time, use clustering method, from a large number of customers with high risk analysis portion of customers, we try to identify the high-risk surrender customer from the large amount of customers.so as to promote the renewal work, improve the company's quality policy, to ensure that the company's long-term business development.
Considering the actual demand of insurance company operation, according to the insurance company’s hardware and software environment, and the insurance’s situation of data, using data mining method, establishes the customer and renewal policy analysis method and model, design the insurance of quality management system, and in the Taiping Life Sichuan Branch successfully applied,through the test and on the line, the system reliable, the effect is very good, to achieve the function, for the control risk and operating costs, improve customer service and service quality to provide a strong and accurate tool and data reference.
本文基于数据挖掘分类方法,分析现有客户的缴费情况,发掘交费概率和交费次数以及交费准时程度的关系,将客户细分为优质客户、准优质客户、忠诚客户、非稳定客户等几类。利用决策树和多元线性回归方法,建立续期客户交率的预测模型,指导续期部门对不同客户采取差异化的催交策略。使保险公司客户管理更有针对性,降低客户保费迟缴、忘缴等风险,并降低管理成本。同时,利用聚类方法,从大量的客户中分析具有较高退保风险的客户,采取相应措施,尽量降低退保事件发生的概率,从而提高续期收费的成功率,提高公司的保单品质,确保公司长期经营发展。
在充分考虑保险公司实际经营需求基础上,将本文所研究的客户续期保单分析模型进行实现,并设计与实现人寿保险品质管理系统。该系统在太平人寿四川分公司顺利应用,系统运行效果好。为该公司保单续期工作的高效开展、控制退保风险和运营成本、提升客户服务和业务品质提供了有力的工具和准确的数据参考。
The insurance industry in China is a rapidly developing industry, business grows quickly to bring more customer information. Along with the insurance company's network speed and server processing capacity continue to improve, based on the data of business intelligence and analysis has gradually become a trend. In recent years, insurance industry in China develops rapidly, the rising volume of customers, the insurance company of the operating costs have increased by a large margin. Insurance company business income is from first-year premium, more important is the renewal premium. Renewal premium has higher stability and continuity characteristics, is the main source of profit, and is a solid foundation for insurance company. At present the way of getting renew premium in insurance company’s very simple,for every customer,they give notice in the same way,it cost a lot and waste a lot. does not favor the insurance risk and cost control.
This thesis try to use data mining technique and classification method to analyze the activities of customer in payment,to discover the relationship between the possibility of payment and how many times they have paid and when they have paid the premium. According to the information of customer,we partition the customer in several group,such as high quality customer,quasi high quality customer,faithful customer and risk customer.Using decision tree and multi-variable regression, we make a prediction model to estimate the possibility of a certain customer’s activity in payment . guide the renewal sectors for different clients to take the difference of expediting strategy. The insurance company customer management more targeted, reduce customer premium late payments, forgot to pay, and reduce the cost of management. At the same time, use clustering method, from a large number of customers with high risk analysis portion of customers, we try to identify the high-risk surrender customer from the large amount of customers.so as to promote the renewal work, improve the company's quality policy, to ensure that the company's long-term business development.
Considering the actual demand of insurance company operation, according to the insurance company’s hardware and software environment, and the insurance’s situation of data, using data mining method, establishes the customer and renewal policy analysis method and model, design the insurance of quality management system, and in the Taiping Life Sichuan Branch successfully applied,through the test and on the line, the system reliable, the effect is very good, to achieve the function, for the control risk and operating costs, improve customer service and service quality to provide a strong and accurate tool and data reference.
引文
[1]焦清平.中国保险业风险形成及治理对策.保险研究,2008,(10):49-53
[2]田中军,续保性是寿险公司持续稳健发展的根本保障.保险研究,2007,(3):36-37
[3]赵迎琳,于泳,杨融,丁文辉.保险业风险管理理论与方法的探讨.保险研究,2003,(7):36-38
[4]商锦博.探索数据挖掘在保险公司中的应用.商场现代化.2007,(19):11-12
[5]陈建成.数据挖掘技术在客户关系管理系统中的应用.电脑与电信.2007,(02):22-24
[6]四川保险行业协会.业绩动态.保险市场动态.2011,(3):5-6
[7]董琴.基于数据仓库的决策支持系统设计与研究.电脑知识与技术,2009,(21):19-21
[8]阮备军,朱扬勇.基于商品分类信息的关联规则聚类.计算机研究与发展,2004,(02):97-105
[9]鲍军鹏,李鹏,刘晓东,沈钧毅.用数据挖掘方法扩充知识库的研究及应用.小型微型计算机系统.2004,(04):142-145
[10]纪希禹.数据挖掘技术应用实例.北京:机械工业出版社.2009,90-130
[11]滕广青,毛英爽.国外数据挖掘应用研究与发展分析.统计研究,2005,(12):68-71
[12]朱浩刚,孙煌鸥,戴伟辉.基于数据挖掘的移动通讯业客户流失管理.计算机工程与应用.2004,(01):215-219
[13]梁循.网络金融信息挖掘导论.北京:北京大学出版社.2008,45-89
[14] Qin liao,Jiabi Zheng.Research on evaluation and optimized structure of managerial completence,Proceedings of the First International Conference on Management Innovation,2007,1(2):761-764
[15] Olivia Parr Rud,朱扬勇.北京:数据挖掘实践.2004,14-50
[16]王玉珍.基于数据挖掘的决策树方法分析.电脑开发与应用,2007,(05):10-13
[17]宋群斌.数据挖掘将为保险业带来新的动力.中国保险报2010,(32)
[18] Tsay,Yuh-JiuanandChiang,Jiunn-Yann.an efficient ethod for mining association rules.Knowledge-Based Systems,Volume18,Issue 2-3,April,2005,99-105
[19]左子叶,朱扬勇.基于数据挖掘聚类技术的信用评分评级.计算机应用与软件.2004,(04):3-5
[20]闰炜,崔杜武,付长龙.基于幂集的关联规则挖掘算法研究.计算机工程与应用,2004,(01):192-193
[21]张劲松.保险公司数据挖掘技术应用探索.商场现代化,2009,(15):108-110
[22]彭宏,郑启伦.基于数据挖掘和范例推理的智能分析决策支持技术综述.计算机工程与应用.2004,(09):l84-187
[23]周丹晨,殷国富,龙红能.CRM环境下面向知识发现的数据分类技术的应用研究.计算机应用与软件,2004,(02):10-12
[24]王新军,胡曼.数据挖掘技术在寿险业交叉销售中的应用.保险研究,2009,(6):92-101
[25]梁循.互联网金融信息智能挖掘基础.北京:北京大学出版社.2007,120-167
[26]陶立慧.数据挖掘技术及在商业决策中的应用研究.商场现代化.2007,(10):15-16
[27] John Wiley,Paolo Giudici.Applied Data Mining:Statistical Methods for Business and Industry.2003,67-71
[28] Grzymala-Busse,JerzyW.A Comparison of Three Strategies to Rule Induction from Data with Numerical Attributes,Electronic Notes in Theoretical ComPuter Seience,Volume82,Issue4,March,2003,1-9
[29]毛国君.数据挖掘原理与算法.北京:清华大学出版社.2005,79-123
[30]马秀红,宋建社,董展飞.数据挖掘中决策树的探讨.计算机工程与应用.2004,(01):185-204
[31]杨静,张楠男,李建,刘延明,梁美红.决策树算法的研究与应用Research and Application of DecisionTree Algorithm.计算机技术与发展,2010,(20):55-57
[32]魏晓云.决策树分类方法研究The Study of Decision Tree ClassifyingMethod in Datamining.计算机系统应用,2007,(9):23-25
[33]舒红平,游志胜,蒋建民.基于信息嫡的决策属性分类挖掘算法及应用.计算机工程与应用.2004,(01):186-189
[34]李如平.数据挖掘中决策树分类算法的研究.东华理工大学学报,2010,(6):192-197
[35]林向阳.数据挖掘中的决策树算法比较研究.中国科技信息,2010,(2):23-27
[36]张德丰,马子龙,梁忠宏.基于聚类和关联规则的挖掘算法.计算机工程与科学,2004,(09):66-68
[37]傅景广,许刚,王裕国.基于遗传算法的聚类分析.计算机工程.2004,(04):126-128
[38]刘书香,卢才武,张志霞.数据挖掘中的客户聚类分析及其算法实现.信息技术.2004,(01):8-10
[39]耿焕同,章曙光,钱权,蔡庆生,王煦法.一种有效的用于范例提取的改进聚类算法.小型微型计算机系统.2004,(03):68-70
[40]夏火松.数据仓库与数据挖掘技术.北京:科学出版社出版地,2004:129-131
[2]田中军,续保性是寿险公司持续稳健发展的根本保障.保险研究,2007,(3):36-37
[3]赵迎琳,于泳,杨融,丁文辉.保险业风险管理理论与方法的探讨.保险研究,2003,(7):36-38
[4]商锦博.探索数据挖掘在保险公司中的应用.商场现代化.2007,(19):11-12
[5]陈建成.数据挖掘技术在客户关系管理系统中的应用.电脑与电信.2007,(02):22-24
[6]四川保险行业协会.业绩动态.保险市场动态.2011,(3):5-6
[7]董琴.基于数据仓库的决策支持系统设计与研究.电脑知识与技术,2009,(21):19-21
[8]阮备军,朱扬勇.基于商品分类信息的关联规则聚类.计算机研究与发展,2004,(02):97-105
[9]鲍军鹏,李鹏,刘晓东,沈钧毅.用数据挖掘方法扩充知识库的研究及应用.小型微型计算机系统.2004,(04):142-145
[10]纪希禹.数据挖掘技术应用实例.北京:机械工业出版社.2009,90-130
[11]滕广青,毛英爽.国外数据挖掘应用研究与发展分析.统计研究,2005,(12):68-71
[12]朱浩刚,孙煌鸥,戴伟辉.基于数据挖掘的移动通讯业客户流失管理.计算机工程与应用.2004,(01):215-219
[13]梁循.网络金融信息挖掘导论.北京:北京大学出版社.2008,45-89
[14] Qin liao,Jiabi Zheng.Research on evaluation and optimized structure of managerial completence,Proceedings of the First International Conference on Management Innovation,2007,1(2):761-764
[15] Olivia Parr Rud,朱扬勇.北京:数据挖掘实践.2004,14-50
[16]王玉珍.基于数据挖掘的决策树方法分析.电脑开发与应用,2007,(05):10-13
[17]宋群斌.数据挖掘将为保险业带来新的动力.中国保险报2010,(32)
[18] Tsay,Yuh-JiuanandChiang,Jiunn-Yann.an efficient ethod for mining association rules.Knowledge-Based Systems,Volume18,Issue 2-3,April,2005,99-105
[19]左子叶,朱扬勇.基于数据挖掘聚类技术的信用评分评级.计算机应用与软件.2004,(04):3-5
[20]闰炜,崔杜武,付长龙.基于幂集的关联规则挖掘算法研究.计算机工程与应用,2004,(01):192-193
[21]张劲松.保险公司数据挖掘技术应用探索.商场现代化,2009,(15):108-110
[22]彭宏,郑启伦.基于数据挖掘和范例推理的智能分析决策支持技术综述.计算机工程与应用.2004,(09):l84-187
[23]周丹晨,殷国富,龙红能.CRM环境下面向知识发现的数据分类技术的应用研究.计算机应用与软件,2004,(02):10-12
[24]王新军,胡曼.数据挖掘技术在寿险业交叉销售中的应用.保险研究,2009,(6):92-101
[25]梁循.互联网金融信息智能挖掘基础.北京:北京大学出版社.2007,120-167
[26]陶立慧.数据挖掘技术及在商业决策中的应用研究.商场现代化.2007,(10):15-16
[27] John Wiley,Paolo Giudici.Applied Data Mining:Statistical Methods for Business and Industry.2003,67-71
[28] Grzymala-Busse,JerzyW.A Comparison of Three Strategies to Rule Induction from Data with Numerical Attributes,Electronic Notes in Theoretical ComPuter Seience,Volume82,Issue4,March,2003,1-9
[29]毛国君.数据挖掘原理与算法.北京:清华大学出版社.2005,79-123
[30]马秀红,宋建社,董展飞.数据挖掘中决策树的探讨.计算机工程与应用.2004,(01):185-204
[31]杨静,张楠男,李建,刘延明,梁美红.决策树算法的研究与应用Research and Application of DecisionTree Algorithm.计算机技术与发展,2010,(20):55-57
[32]魏晓云.决策树分类方法研究The Study of Decision Tree ClassifyingMethod in Datamining.计算机系统应用,2007,(9):23-25
[33]舒红平,游志胜,蒋建民.基于信息嫡的决策属性分类挖掘算法及应用.计算机工程与应用.2004,(01):186-189
[34]李如平.数据挖掘中决策树分类算法的研究.东华理工大学学报,2010,(6):192-197
[35]林向阳.数据挖掘中的决策树算法比较研究.中国科技信息,2010,(2):23-27
[36]张德丰,马子龙,梁忠宏.基于聚类和关联规则的挖掘算法.计算机工程与科学,2004,(09):66-68
[37]傅景广,许刚,王裕国.基于遗传算法的聚类分析.计算机工程.2004,(04):126-128
[38]刘书香,卢才武,张志霞.数据挖掘中的客户聚类分析及其算法实现.信息技术.2004,(01):8-10
[39]耿焕同,章曙光,钱权,蔡庆生,王煦法.一种有效的用于范例提取的改进聚类算法.小型微型计算机系统.2004,(03):68-70
[40]夏火松.数据仓库与数据挖掘技术.北京:科学出版社出版地,2004:129-131