Research On The Application Of Discriminating The Default State Of Small Business

Small businesses play an important role in the national economy and are the main force driving my country's economic progress.As an important participant in the financial market,small businesses are the main target for banks to issue loans.However,in the long-term development,due to their weak ability to cope with financial risks,small businesses do not have enough capital to repay loans to banks and default.Therefore,how to realize the accurate judgment of the default status of small enterprises has become an urgent problem to be solved.Aiming at this problem,this article focuses on designing a set of accurate default judgment solutions,and through a mini-case analysis of Dalian Parkland Co.,Ltd.,it shows that the company has three problems: insufficient profitability,insufficient experience,and lack of liquidity.It also proposes corresponding solutions to these three problems,such as changing business models,improving credit risk prevention awareness and experience,and expanding market business.This article studies the application of judging the default status of small enterprises.The first chapter of this article is the introduction,the second chapter is the construction of the default judgment model,the third chapter is the empirical analysis of the judgment of the default status of small enterprises,the fourth chapter is the judgment and analysis of the default status of Dalian Parkland Co.,Ltd.,and the conclusion is the final.The research focus of this paper is twofold: The first is the best division of samples.The training set and the test set are different when the sample is divided in different ways,and the established default state discrimination model is different,which leads to the different discriminant accuracy of the model,so it is adopted What kind of sample division method can maximize the discrimination accuracy of the model and realize the best division of samples is an urgent problem to be solved.The second is the optimization of the standard deviation parameters of the Probability Neural Network(PNN)discriminant model.The value of the standard deviation parameter is different,so there must be an optimal value of the standard deviation parameter to ensure the maximum discriminant accuracy of the model.There are two innovations in this paper: One is to use the maximum AUC as the objective function to determine the accuracy of the default of the probabilistic neural network,and inversely infer the optimal number of clusters;under the premise of the optimal number of clusters,gather customers with small Euclidean distance In a category,the clustering results are arranged in descending order of the number of clustered customers,and the customers in the first few categories whose number of clustered customers account for 80% of the total number of customers are regarded as training customers,and the remaining few The customers in the category are used as test customers.The training set and the test set obtained by this method are divided,considering the similarity of the data distribution and the influence of the similarity of the data distribution on the model,to achieve the best division of the sample.The best sample division method proposed in this research improves the discrimination accuracy of the probabilistic neural network model.The second is to use 0.1 as the step size,in the(0,1)interval,take the value of the standard deviation parameter of the probabilistic neural network,and use the AUC value of the default judgment accuracy of the probabilistic neural network to maximize the optimal standard deviation parameter to avoid The disadvantages of the existing research of randomly selecting the value of the standard deviation parameter are improved,and the accuracy of default judgment based on the probabilistic neural network model is improved.There are two findings in this article: First,19 indicators such as "current ratio","earnings before interest and taxes","overspeed ratio" and other 19 indicators have an important influence on the judgment of the default status of small enterprises.The second is the number of five indicators: "net cash flow from operating activities","corporate legal disputes","number of breaches of contracts between enterprises","cost-profit ratio",and "current-to-liability ratio before interest and tax" Although they only account for 26.32% of all indicators,their importance accounts for 61.83% of all default indicators.