基于Logistics-ANN的小企业信用评估方法及其应用研究

发布时间：2018-04-28 16:57

本文选题：主成分分析 + Logistics回归　；参考：《哈尔滨工业大学》2012年硕士论文

【摘要】：从上个世纪20年代以来，关于小企业信用评估的探索就已经成为国外信用评估研究的热点，相关学者立足于不同的角度和方法对小企业信用评估进行了深入全面的研究。经过多年的发展，在形成了一定理论研究成果的同时，该方法的应用也极大的推进了当地经济的发展和提高。在国外的研究中，小企业具有一套适合自身特征的信用评估方法和指标体系。但是目前我国在小企业的信用评估方面尚处于起步阶段，大部分研究是将小企业与中型企业混合在一起探讨信用评估方法，这与小企业自身特征不符，而国内专门针对小企业的文献研究少之又少，尚未形成一套完善的小企业信用评估体系。鉴于目前国内对小企业信用评估的研究现状，本文在对国内外相关理论进行梳理的基础之上，从小企业自身特征出发，通过对小企业信用数据及指标的特点进行分析，并结合现有的研究方法及其应用成果，得出了现有研究方法本身的不足之处以及在小企业信用评估应用方面存在的缺陷，由此构建了由Logistics分析和神经网络ANN分析共同组成的小企业信用评估方法，即首先针对小企业的指标特点引入Logistics方法进行指标筛选，随后将筛选出来的显著性变量作为神经网络模型的输入变量进行信用评估，并对该方法在小企业信用评估方面的应用进行了深入研究，最后将其应用结果与现存的小企业信用评估方法进行对比，，得出了该方法具有判别精度高、简单易行等特点，进而结合该结果对未来的小企业信用评估研究提出了相关的结论和建议。
[Abstract]:Since the 1920s, the research on the credit evaluation of small enterprises has become a hot topic in foreign countries, and some scholars have made a thorough and comprehensive study on the credit evaluation of small enterprises from different angles and methods. After years of development, some theoretical research results have been formed, and the application of this method has greatly promoted the development and improvement of local economy. In foreign research, small enterprises have a set of credit evaluation methods and index system suitable for their own characteristics. However, at present, the credit evaluation of small enterprises in China is still in its infancy, most of the research is to mix small and medium-sized enterprises to discuss the methods of credit evaluation, which is not consistent with the characteristics of small enterprises. However, there is very little literature research on small enterprises in China, which has not yet formed a perfect credit evaluation system for small enterprises. In view of the current domestic research status of small business credit evaluation, this paper, on the basis of combing the relevant theories at home and abroad, starts from the characteristics of small enterprises, through the analysis of the characteristics of small business credit data and indicators. Combined with the existing research methods and their application results, the shortcomings of the existing research methods and the shortcomings in the application of small enterprises credit evaluation are obtained. Based on this, a credit evaluation method for small enterprises is constructed, which is composed of Logistics analysis and neural network ANN analysis. Firstly, the Logistics method is introduced to screen the indexes according to the characteristics of small enterprises. Then the salient variables are used as the input variables of the neural network model to evaluate the credit, and the application of this method in the small enterprise credit evaluation is studied deeply. Finally, the application results are compared with the existing small enterprise credit evaluation method, and it is concluded that the method has the characteristics of high accuracy, simplicity and so on. Based on the results, some conclusions and suggestions are put forward for the future research on credit evaluation of small enterprises.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F275;F276.3;F832.4

【参考文献】