基于PCA-LSFSVM的供应链金融信用风险评估模型研究

发布时间：2018-01-24 23:43

本文关键词： 供应链金融信用风险评估主成分分析法最小二乘模糊支持向量机　出处：《华南理工大学》2014年硕士论文　论文类型：学位论文

【摘要】：供应链金融服务起源于20世纪90年代，最开始是由国内外的先进物流企业和商业银行开展的。国内外一些先进的物流企业及银行适时地推出了多样化的供应链金融服务，以帮助企业实现对资金流的管理。近年来供应链金融己成为许多物流企业、金融机构关注的焦点。在国内，供应链金融业务是针对供应链的业务结构、交易方式及运作特点而设计的，以供应链中最具有信用实力的大企业为核心，，围绕与之在生产过程、销售过程中开展业务的多家中小企业，将整个供应链链条涉及到的上下游企业连起来，整体考察供应链的信用状况，更好地为中小企业提供金融服务，从而促进整个供应链价值的增长。过去商业银行对中小融资企业信用风险的评估主要是着力于单个中小融资企业本身，分析其财务数据，关注其抵押物质等，而在供应链金融融资模式中，对中小融资企业风险的认识和评估则换了一个新的视角。把供应链融资模式中的核心企业也作为考察的主体，以及整个供应链链条的稳定作为参考因素。本文以供应链金融为视角研究商业银行对中小融资企业信用风险的评估，首先通过分析供应链金融融资模式的特点，构建基于供应链金融的信用风险评价指标体系；接着采用主成分分析法对指标数据进行降维处理，简化数据结构提高分类效率；对比分析现有信用风险评价方法的优缺点，确定用支持向量机SVM方法进行评估，具体选择了带模糊隶属度的模糊支持向量机并引入可变罚因子对其算法进行改进，在此基础上建立了供应链金融信用风险评估模型。模型的求解过程中，引入二次损失函数将复杂的二次规划问题转化为求解线性方程，求出最优分类函数。最后通过实证研究，将基于主成分分分析与最小二乘模糊支持向量机（PCA-LSFSVM）的供应链金融信用风险评估模型与基于二分类Logistic回归的供应链金融信用评估模型进行对比分析，证实了基于PCA-LSFVM的供应链金融信用风险评估模型性能优越，能有效地解决供应链金融的信用评估问题，从而为商业银行供应链金融信用风险评估提供科学合理的有效工具，有着重要的理论价值和现实意义。
[Abstract]:Supply chain financial services originated in 1990s. At the beginning, it was carried out by advanced logistics enterprises and commercial banks at home and abroad. Some advanced logistics enterprises and banks at home and abroad launched diversified supply chain financial services in good time. In recent years, supply chain finance has become the focus of many logistics enterprises and financial institutions. In China, supply chain financial business is aimed at the business structure of supply chain. The transaction mode and operation characteristics of the design, the supply chain with the most credit strength of the large enterprises as the core, around the production process, sales process business with a number of small and medium-sized enterprises. Connecting the upstream and downstream enterprises involved in the whole supply chain, the credit condition of the whole supply chain is investigated, and the financial service is provided to the small and medium-sized enterprises better, thus promoting the growth of the value of the whole supply chain. In the past, commercial banks mainly focused on the credit risk assessment of small and medium-sized financing enterprises, analyzed their financial data, paid attention to their mortgage materials, and so on, but in the financing mode of supply chain finance. The cognition and evaluation of the risk of the medium and small financing enterprises have changed a new angle of view. The core enterprises in the financing mode of supply chain are also taken as the main body of investigation, and the stability of the whole supply chain is taken as the reference factor. This paper studies the credit risk assessment of small and medium-sized financing enterprises by commercial banks from the perspective of supply chain finance. Firstly, it analyzes the characteristics of the supply chain financial financing model. Constructing credit risk evaluation index system based on supply chain finance; Then the principal component analysis (PCA) is used to reduce the dimension of the index data and simplify the data structure to improve the classification efficiency. Comparing and analyzing the advantages and disadvantages of the existing credit risk evaluation methods, the support vector machine (SVM) SVM method is used to evaluate. The fuzzy support vector machine with fuzzy membership is selected and the variable penalty factor is introduced to improve the algorithm. On this basis, the credit risk assessment model of supply chain finance is established. In the process of solving the model, the quadratic loss function is introduced to transform the complex quadratic programming problem into solving the linear equation, and the optimal classification function is obtained. Finally, the empirical study is carried out. PCA-LSFSVM based on principal component analysis (PCA) and least square fuzzy support vector machine (LSSVM) is proposed. The model of supply chain financial credit risk assessment is compared with the supply chain financial credit evaluation model based on two-classification Logistic regression. It is proved that the credit risk assessment model of supply chain finance based on PCA-LSFVM has superior performance and can effectively solve the credit evaluation problem of supply chain finance. Therefore, it is of great theoretical and practical significance to provide a scientific and effective tool for the credit risk assessment of the supply chain finance of commercial banks.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832;F224

【参考文献】