导　　师： 傅红卓

学科专业： 070103

授予学位： 硕士

作　　者： ;

机构地区： 华南理工大学

摘　　要： 本文以银行风险管理的相关理论为指导，对集体经济实体贷款各方面进行研究，分析在现行信贷风险评价理论和方法下集体经济实体信用评价的情况。通过深入调查研究，提出了集体经济实体信贷风险可能的变化趋势。结合目前我国经济发展的实际情况与需要，特别是集体经济发展的现状，以信贷风险管理中的Zeta和Z评分模型及CreditMetrics模型等为理论基础，结合集体经济实体信贷的实际案例，对集体经济实体信贷风险分析和度量进行了剖析，构建集体经济实体信贷风险预测模型，为银行信贷风险管理领域的研究提供参考。 主要思路是：以我国集体经济业务为切入点，针对集体经济信贷业务发展中存在的违约风险，应用数据挖掘技术于违约的主要源头——信用风险评估，通过建立基于神经网络算法的信用评估模型，对申请信贷业务的借款人信用状况进行评估，预测出借款人可能的违约状况，从而从贷款审批源头上控制风险。 主要内容：第一部分对集体经济实体信贷的发展状况和特点进行比较分析，进而得出集体经济实体信贷风险的多方面成因。第二部分对数据挖掘理论的原理、过程，以及人工神经网络主要算法的原理和应用进行研究。第三部分在对集体经济信用风险成因深入研究的基础上，研究如何将人工神经网络算法应用于信用评估的建模过程，通过建立BP神经网络模型，找出集体经济实体信息指标与信用评估之间隐含的规律。根据模型的结果得到各信息指标对于不良信用的影响程度，从而确定各指标的权重并建立集体经济实体信用评价模型，将数据挖掘结果量化为具体的评分模型。 本文结论：1、在当前集体经济实体贷款蓬勃发展的情况下，对集体经济实体信用风险评价有必要采用新方法；2、运用BP神经网络建立的集体经济实体信用风险评价模型适合集体经济实际情况。 本文模型的数据源于广州本地集体经济组织信息，由于数据挖掘最大的优点就是“原生性”，建立的模型立足于本土，与国内信贷市场相吻合，有极强的适应性和原生性。此外该模型的建模方法、数据处理方式也可以适用于其它类贷款的信用评估，具有一定的普遍适用意义。 The paper bases on the theory of banking risk mangagement. In addition,the empirical research of village-level enterprises loans is used by comparison of current credit-debit risk evaluation theory and methods. Applying mathematical statistics operational research and computer simulation formed the optimal model on the credit risks of village-level economil entity. The paper with the bank credit risk management in of Zeta and Z grade point model and Credit Metrics the market risk generous character system model etc.for theories foundation,pass to combine village-level economil entity actual case,technique and method draw lessons from foreign advanced science,combine domestic economy and finance of actual,to village-level economil entity risk analysis and generous character,put forward to build suitable Credit Risk Forecast Model of village-level economil entity,for village-level economil entity credit project the research of the risk management realm provide a reference. The main idea of this paper is that based on the Credit Loan business of village-level economil entity,for the exit of default risk in business development,it will use of technology of Data Mining to assess the credit risk and build a credit assess model which is based on the algorithm of Neural Network. And this model can predict the borrower's possible statement of default and bad debt rate. It will control the original risk. The first part of this paper analyses the deveiopment and of Credit Risk of village-level economil entity business and then gets to the various causes of village-level economil entity Credit Risk Assessment Methodology. The second part of this paper studies the principles of Data Mining theory and process. And then this paper proceeds the building of credit assess models. It finds the disciplines between the information and credit assessment by the comparison of BP Neural Networks. The final model can predict the possible numbers of new borrower and reduce the risk of default. At last, according to the results of model it will create a Credit Evaluation System which can quantize the risk of default and improve the efficiency of loan approval. The model of credit risk assessment on listed village-level economil entity is effective. The date which comes from local village-level economil entity of guangzhou have good 'original' advantage. What's more, the Method of Modeling can be used on the other assessment of loan and has a meaning of universal application meaning.

关 键 词： 预测模型 信贷风险 风险评价 集体经济实体

分 类 号： [F224 F832.4 F121.22]

领　　域： [经济管理] [经济管理] [经济管理]