导　　师： 苏越良

学科专业： 1201

授予学位： 硕士

作　　者： ;

机构地区： 华南理工大学

摘　　要： 商业银行是目前我国社会进行直接融资的最大来源。商业银行为了现代社会做出了巨大的贡献，但社会过渡依赖银行信贷导致中国商业银行放贷压力过大、风险高度集中，很容易引发银行系统性的金融风险。商业银行存在的诸多业务风险中，影响最大的就是信用风险，主要指贷款的客户信用严重缺失，尤其是中小企业的容错性远远比不上大型企业，信用风险问题更为严重。这样一方面给商业银行造成巨额损失，削弱了银行的盈利能力；另一方面，还加重了银行的不良资产负担，阻碍了银行的进一步放贷，甚至会降低银行的信用状况，引发银行的破产。因此商业银行中针对中小企业的信用风险管理也成为银行风险管理中最关键、最具挑战性的工作。而针对中小企业信用风险管理最有效的方法就是对深入了解信贷企业的信用状况，对其进行信用评级，并依据信用级别采取相应的信贷策略。 本文首先回顾了中小企业信用评价的研究现状，并根据商业银行对中小企业信用评价的特点，构建一套具有完备性的中小企业信用评价指标体系，以综合反映中小企业的信用状况；接着，采用主成分分析方法对指标数据进行降维处理，以提取几个综合的主成分来代替原始指标体系，实现数据结构的简化并减少数据的相关关系，为后续模型的输入提供数据支持；在此基础上利用LSSVM技术对中小企业信用进行评价，通过此模型来解决指标数据复杂性及非线性问题，以寻找最优的回归函数。通过实证分析表明，基于主成分分析和LSSVM信用评价性能优于BP神经网络。 本文运用主成分分析和LSSVM模型相结合的方法对中小企业的信用进行评价，不仅为商业银行建立起科学的商业银行内部中小企业信用评价体系，而且为其更准确的量化和分析中小企业的信用状况提供了理论依据和决策工具。这对商业银行强化信用风险监管，防范和化解信用风险，提高银行的资产质量有着极大的理论意义和实践价值。 Commercial Banks is the biggest source of direct financing in our country. CommercialBanks make great contribution to modern society, but the social transition rely on the credit ofbank, which leads to China's commercial Banks with more stress, higher risk, so that it cancause the systemic financial risk of banks. There are many risks in Commercial Banks, but thecredit risk is the biggest one, mainly refers to the customer credit default, especially the faulttolerance of small-medium enterprises is less than the large-scale enterprise, and its credit riskproblem is more serious. On the one hand, it can cause huge losses, and weak the Banks'profitability; On the other hand, it also can increase burden of the bad assets, even reduce thecredit conditions of banks, and trigger a bank failure. Therefore, the credit risk managementof the small-medium enterprises is the in the key and the most challenging job for commercialbanks. For the credit risk management of the small-medium enterprises, the most effectiveway is understanding the credit conditions of enterprises, and rating credit, then usingeffective credit strategy based on the credit level. First, this paper reviews research on credit evaluation of the small-medium enterprises,and according to the characteristics of credit evaluation in small-medium enterprises, weconstruct a credit evaluation index system of the small-medium enterprises completely, whichreflects the credit conditions more comprehensively; Then, we use the principal componentanalysis method to processing the index data, and extracting some comprehensive principalcomponents to replace the original index system, then simplifies the data structure andreduces data correlation for the input data of the subsequent model; On the basis of usingLSSVM to evaluate the credit of small-medium enterprises, through this model to solvecomplexity and nonlinear problem of the data, so that we can find the most optimumregression function. Through empirical analysis shows that, the credit evaluation performancebased on principal component analysis and LSSVM is better than the BP neural network. This paper by using principal component analysis and LSSVM model to evaluate thecredit of the small-medium enterprises, not only established a scientific credit evaluationsystem of small-medium enterprise for commercial Banks, but also it have more accuratequantitative and analysis for the credit conditions of the small-medium enterprises, which canprovide the theoretical and tool for decision making. SO it can strengthen the credit risksupervision for commercial banks, and guard against and dissolve the credit risk, thenimprove the quality of the assets of the bank, which has a great theoretical significance and practical value for c ommercial banks.

关 键 词： 中小企业 信用评价 主成分分析 最小二乘支持向量机

分 类 号： [F832.4 F276.3 F224]

领　　域： [经济管理] [经济管理] [经济管理] [经济管理]