建立基于神经网络和正交设计的个人信用评分体系初探-管理科学与工程专业论文.docxVIP

第11页 第11页 武汉科技大学硕士学位论文 Abstract It developed lately of Chinese consumer credit evaluation．In the early era，consumer credit was evaluated based on judgment．At present，the production of research of financial and theory field is macro-based and qualitative．Application researches in micro—level ale less much more． ha practice，we must popularize and strengthen the quantitative model in the operation of commercial banks．It was used at large that the credit scoring in practical operation．The basic idea is confirm some key factors of breach probability before add them up with mathematical method，then comparative the power with the figure confirmed beforehand and decide if accept the application of the borrower or not and the quantum of the loan．Chinese credit scoring system was used for reference from foreign with little adjusted． This paper will put forward some scientific ideas of constitute effective and economical credit scoring model．And make this model magnificent in theory and practice． It is divided 3 main part of the paper．The 1“part showed mainly circumstance of consumer credit of China and abroad．The 2Ⅱ。part illuminated the basic theory of the ANN(Artificial Neural Netwolk)and orthogonal design．In the 3“part，it was proposed based on the explain of fore context that the main constitution thinking of credit model，and explained the insufficiency of the paper and the prospect of the credit scodng model． Key Words：Consumer，Credit Scoring，ANN，BP Neural Network，Orthogonal Design 武汉科技大学硕士学位论文 武汉科技大学硕士学位论文 第1页 前言 随着我国加入wTO，经济开放程度越来越大，各种信贷活动将越来越多，消费者参加 的信贷活动将进一步扩大，消费者信用问题也将更加突出，为保证我国金融秩序的稳定， 结合各种新技术、新思想提出行之有效的消费者信用评价体系，对为决策者提供科学的决 策依据，及时识别消费者信用风险，具有十分重要的意义。 从目前来看，我国金融界和理论界对消费信贷风险管理的研究成果大多是宏观上的、 定性分析为主，微观层次上的应用研究很少，定量分析模型在实际业务中的应用还须加强 和推广。我国消费信贷起步较晚，早期一般用判断式信用评分，银行通过借款申请人在申 请书上填写的内容来判断贷与不贷，贷款多少，长短期限等，是一种经验式的、人为因素 较多的判断方式。至于个人信用评分，在我国商业银行中，有些没有做过这方面的工作， 有些还刚刚起步。我国现有的消费者评分体系主要是借鉴国外银行的消费者信用评分体 系，在结合我国实际情况调整后形成本系统的评分体系。由保障支持、经济支持、稳定情 况和消费者背景四部分内容组成，采用百分制。消费者信用评分体系的推出，使银行能够 更快