Credit Evaluation Index System Research Of Science And Technology Enterprises Based On GA-BP Algorithm

Science and technology enterprises have "three high" characteristics--high-tech, high-growth and high-risk. Under the background of the rapid development of global technology innovation and the "hedge" feature of China’s existing financial system, "financing difficulties" is a long-term problem facing in its process of continuous development. In our country, the commercial bank loan is the principal means of solving the financing difficulties of science and technology enterprises, is also the primary way of indirect financing. However, because the quantity of science and technology enterprises is large, the history of development is short, the market risk is relatively large, operating profit is unstable, the asset owned by the enterprises which can be clearly evaluated and loaned is less and etc., the credit level of the enterprises is generally not high, so it is very hard to get through the loan qualification of commercial banks.Taking science and technology enterprises in our country as the research object, this paper first introduces the basic theory of evaluation index system and the credit rating model, and compares the artificial neural network with common rating models. Then this paper expounds the setting principles of credit evaluation for science and technology enterprises and puts forward the credit evaluation index of science and technology enterprises, and then uses AHP and expert scoring method to conduct an in-depth analysis and screening of the indicators, based on this, this paper establishes a scientific, reasonable and effective credit evaluation index system of science and technology enterprises, and through the analysis of the industry characteristics and the credit demand of science and technology enterprises, this paper carries on an empirical study based on 25 science and technology enterprises in Tianjin-a credit rating model in connection with science and technology enterprises in our country based on the BP neural network optimized by genetic algorithm. On the basis of inheriting such superiorities of traditional BP neural network as self-learning, adaptive and nonlinear mapping ability, based on GA-BP algorithm neural network model overcomes the defects which traditional BP algorithm mostly has, such as falling into local minimum easily, slower convergence speed and hard determination of model structure, reduces the output error effectively, improves the performance of the model and the generalization ability of BP neural network. Through data validation, this paper finally shows that based on GA-BP algorithm neural network model is completely effective and feasible for the establishment of credit evaluation index system of science and technology enterprises. Results of this study have great significance in practice for further research in credit evaluation index system of science and technology enterprises and important implications for solving the financing difficulty of science and technology enterprises.