Research On Financial Warning Of Automotive Industry Based On Adaptive Neural-fuzzy Inference System

Manufacturing industry is the source of national wealth, while the automobile manufacturing industry is the priority among priorities among them. Although China's automobile manufacturing scale and number of enterprises are growing rapidly, the entire industry risk resistance ability is very weak, easily affected by the overall economic environment and national policy, which caused the serious financial crisis. Therefore, the construction of financial early warning model of automobile manufacturing industry, has very important significance to enhance the ability of automobile manufacturing industry to deal with the financial crisis.This paper firstly introduces the background, theoretical and practical significance of the research, on the basis of the domestic and foreign scholars about the financial early warning research, the research methods, ideas and innovation are proposed. Then introduces the concept about enterprise financial early warning, ANFIS and the 3 kinds of clustering algorithm and the method of combination of ANFIS and clustering algorithm. Then, based on the assumption that the reference of previous research, the impact factors of automobile manufacturing industry financial early warning, including the financial and non-financial aspects, are clarified. Seven factors, financial factors have cash flow and short-term debt paying ability, operation ability, growth ability, profitability, non financial factors is consisted of ownership structure and corporate governance. The financial early-warning index system is the basis of early warning model construction, it is one of the important factors directly affect the effect of the early warning. In order to more comprehensively, objectively and truly reflect the financial status of automobile manufacturing industry, the 26 financial indexes of debt paying ability, growth ability, profitability, cash flow ability and operation ability and so on to measure the financial factors. Index including the proportion of the first shareholder 7 non-financial indicators are also used to construct the financial early-warning index system.In the part of empirical research, this paper selected 54 automobile manufacturing and parts manufacturing listed companies to construct required samples set from the Shenzhen and Shanghai stock exchange in A shares, the time span of the samples set is11 years(2003-2013). A third of the samples set are financial crisis enterprises,and the other two thirds matched non financial crisis of enterprises. Processing significant test and principal component analysis for all the samples of the 33 financial and non-financial indicators,deleting some financial and non-financial index whose correlation is not obvious, after the dimension reduction of financial index, finally got 5 financial indicators and 4 non-financial indicators to build China's automobile manufacturing industry listing Corporation financial early warning index system. By employing the index system and the sample data, using the method of ANFIS clustering algorithm based on the construction of early warning model for automobile manufacturing enterprises and BP neural network, SVM(support vector machine) and Logit regression of the three kinds of early warning model, through the model test and the comparative analysis, we can draw the following conclusions: industry listing Corporation financial early warning the model is better than the other models making clustering algorithm based on ANFIS bus, in addition to the model and the BP neural network model and SVM(support vector machine) and other artificial intelligence model, significantly better than the non Logit regression model of artificial intelligence.Through the above research, this paper aimed at the automotive manufacturing industry listing Corporation financial early warning practice put forward two suggestions:(1) construction of the automobile manufacturing industry financial early warning index system should delete some interference index.(2) while selecting the modeling tool, artificial intelligence modeling tool should be distributed priority.