基于数据挖掘的银行对私客户行为研究

发布时间：2018-03-05 11:20

本文选题：数据挖掘　切入点：聚类分析　出处：《浙江工业大学》2014年硕士论文　论文类型：学位论文

【摘要】：近两年来浙江经济形势未有明显改观,银行资产质量持续下行,不良率高企,严重侵蚀银行利润,银行亟待提升对客户的风险管理水平。当前银行对私客户的个人贷款业务和信用卡业务量在全行资产业务量中占比近为四成,面临了较大的信用风险、市场风险和操作风险,机构涉及面广,对整体的资产质量影响较大。传统的风险管理办法基于常规的思路,且多为事后管理性质,使用的技术手段相对落后,对风险点的监控也属于零散或者是无规律,缺乏系统性管理,已不能满足日益迅猛发展的银行业务。本文在对数据挖掘相关理论大量文献研究综述的基础上,将对私客户在上述两种业务中有意无意表达出的行为进行分析,通过数据清理和转化,为数据挖掘提供规范化数据,并使用IBM的SPSS Modeler数据挖掘工具,通过聚类和决策树等建模方法,对客户行为进行了深度分析和挖掘,试图找出形成风险的影响因素,分析偏离正常值的异常信息,形成有参考价值的模型用于预测未来的风险以及监控当前的风险。最后依据数据的可获取性,以及对数据挖掘形成模型的研究后,提出对对私客户的行为监测指标,以供对口条线管理部门参考,提高管理层决策水平,为信贷政策提供准确的数据基础,并提供科学依据,制定业务发展目标。
[Abstract]:In the past two years, the economic situation in Zhejiang has not improved significantly. The quality of bank assets has continued to decline and the rate of failure has remained high, which has seriously eroded bank profits. Banks urgently need to improve the level of risk management for customers. At present, the amount of personal loans and credit card business of banks to private customers accounts for nearly 40% of the bank's total assets business, and they are faced with greater credit risk, market risk and operational risk. The traditional risk management methods are based on conventional thinking, and most of them are ex post management, and the technical means used are relatively backward. The monitoring of risk points is also scattered or irregular, lack of systematic management, can not meet the rapid development of banking business. This paper on the basis of a large number of data mining related theory literature review, This paper will analyze the behavior expressed by private customers in the above two kinds of business, provide standardized data for data mining through data cleaning and transformation, and use the SPSS Modeler data mining tool of IBM. Through clustering and decision tree modeling methods, this paper makes a deep analysis and mining of customer behavior, tries to find out the influencing factors of risk, and analyzes the abnormal information that deviates from the normal value. A model with reference value is formed to predict future risks and monitor current risks. Finally, according to the availability of data and the research of data mining model, the behavior monitoring index for private customers is proposed. In order to provide reference for the corresponding line management department, improve the management decision-making level, provide accurate data basis for credit policy, and provide scientific basis to formulate business development goals.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP311.13;F832.2

【参考文献】