基于GA-BP算法的科技型企业信贷评价指标体系研究

发布时间：2018-03-25 15:14

本文选题：科技型企业　切入点：信贷评价指标体系　出处：《天津财经大学》2014年硕士论文

【摘要】：科技型企业具有高科技性、高成长性、高风险性的“三高”特点,在全球科技创新飞速发展和我国现有金融体系“避险”特性的背景下,“融资难”问题是其不断发展壮大过程中所面临的长期问题。在我国,商业银行贷款是解决科技型企业融资困难的主要手段,也是首要的间接融资方式。但由于科技型企业数量多,发展历史短,市场风险相对大,经营利润不稳定,企业所拥有的可以明确进行评估和抵押的资产少等特点,导致企业的信用级别普遍不高,因此很难通过商业银行的贷款资格审查。本文以我国科技型企业为研究对象,首先介绍了信贷评价指标体系基本理论和信贷评价模型,并将人工神经网络模型与常见评价模型进行比较分析。随后阐述了科技型企业信贷评价指标的设置原则并提出本文选取的评价指标,接着运用层次分析法和专家打分法对指标进行深入分析并赋予权重,以此建立一套科学、合理、有效的科技型企业信贷评价指标体系,并通过分析科技型企业的产业特征和信贷需求,最终以天津市25家科技型企业为例进行了实证研究：基于遗传算法优化的BP神经网络建立了一个针对我国科技型企业的信贷评价模型。基于GA-BP算法的神经网络模型克服了传统BP算法容易陷入局部最小收敛速度较慢、模型结构不易确定等缺陷,同时,在继承传统BP神经网络的自学习、自适应和非线性映射能力的基础上,有效降低了输出结果的误差,改善了模型的性能,提高了BP神经网络的泛化能力。最后通过数据的验证说明了基于GA-BP算法的神经网络模型对于科技型企业信贷评价指标体系的建立是完全有效并可行的。本文的研究成果对进一步在实践中探索科技型企业信贷评价指标体系和解决科技型企业融资难的问题具有重要意义。
[Abstract]:Science and technology enterprises have the characteristics of high technology, high growth and high risk. In the context of the rapid development of global scientific and technological innovation and the "safe haven" characteristics of our existing financial system, the problem of "financing difficulty" is a long-term problem in the process of its continuous development and expansion. Commercial bank loans are the main means to solve the financing difficulties of sci-tech enterprises and the primary indirect financing methods. However, due to the large number of sci-tech enterprises, short history of development, relatively large market risks and unstable operating profits, The characteristics that enterprises have, such as less assets that can be clearly assessed and mortgaged, lead to a general low credit level of enterprises, so it is difficult to pass the loan qualification examination of commercial banks. This paper takes our country's science and technology enterprises as the research object. First of all, it introduces the basic theory of credit evaluation index system and credit evaluation model. The paper compares the artificial neural network model with the common evaluation model, then expounds the setting principle of the credit evaluation index of the science and technology enterprise and puts forward the evaluation index selected in this paper. Then it uses the analytic hierarchy process and the expert scoring method to carry on the thorough analysis to the index and endows the weight, thus establishes a set of scientific, reasonable, effective science and technology enterprise credit appraisal index system. And through the analysis of the industrial characteristics and credit demand of science and technology enterprises, Finally, 25 enterprises in Tianjin are taken as an example. BP neural network based on genetic algorithm optimization is used to establish a credit evaluation model for Chinese science and technology enterprises. Neural network model based on GA-BP algorithm. It overcomes the slow convergence speed of traditional BP algorithm, which is easy to fall into local minimum convergence. The model structure is difficult to determine and so on. At the same time, on the basis of inheriting the self-learning, adaptive and nonlinear mapping ability of the traditional BP neural network, the output error is effectively reduced, and the performance of the model is improved. The generalization ability of BP neural network is improved. Finally, the neural network model based on GA-BP algorithm is proved to be effective and feasible for the establishment of credit evaluation index system of science and technology enterprises. The research results are of great significance to further explore the credit evaluation index system of science and technology enterprises in practice and to solve the problem of financing difficulties of science and technology enterprises.
【学位授予单位】：天津财经大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F275;F832.4

【参考文献】