| In recent years,a number of real estate enterprises,such as Blu ray Development,Xiexin Yuanchuang,Huaxia Happiness,have successively defaulted,and the credit risk of the real estate industry has aroused widespread concern.With the continuous pressure on the real estate industry,bond issuing enterprises in the real estate industry have been downgraded in succession.Against the background of the continuous downgrade of corporate credit ratings,a large number of real estate enterprises can not bear the pressure of the market and public opinion,and take the initiative to apply for the cancellation of credit ratings,leading to the intensification of information asymmetry between bond issuing enterprises and investors.Whether it is the event of default of real estate enterprises or the termination of credit rating of real estate enterprises,the solvency and financing capacity of real estate enterprises are weakened,and real estate enterprises are faced with the severity of high credit risk.Therefore,the market,government and other relevant departments have an urgent need for credit risk warning of real estate bond issuing enterprises.Therefore,it is of great practical significance to study the credit risk warning of real estate bond issuing enterprises.First of all,this paper combs the relevant literature at home and abroad,analyzes the characteristics of credit risk in the real estate industry,and summarizes the factors affecting the credit risk of bond issuing enterprises;Secondly,it introduces the support vector machine(SVM)model,genetic algorithm(GA)and the method of constructing the credit risk early warning model of bond issuing enterprises based on GA-SVM;Thirdly,early warning indicators are preliminarily selected,and relevant coefficient method and random tree method are used to screen indicators according to the importance of characteristics,so as to further establish the credit risk early warning indicator system of real estate bond issuing enterprises;Then,this paper takes bond issuing enterprises in the real estate industry from 2018 to 2021 as the research sample,selects their credit default or credit rating of Baa3/BBB-and below as the credit risk characteristics,preprocesses and filters the sample index data,and uses SMOTEENN sampling method and GA to improve the accuracy of the enterprise credit risk early warning model.Different optimization algorithms are used to optimize the parameters of the SVM model,and the accuracy rate,recall rate and F1 value are used as evaluation criteria to evaluate the model performance.The research conclusions of this paper are as follows:(1)The results of feature selection of credit risk influencing factors by correlation coefficient method and random tree classifier show that the credit characteristics of 14 indicators are outstanding and should be included in the credit risk indicator system;The importance of the credit characteristics of cash ratio,net interest rate of total assets and net cash content of operating income are listed in the top three,indicating that the financial situation of real estate enterprises is the primary factor of credit risk of debt issuing enterprises.The growth rate of per capita income of residents has a significant impact on the credit risk of real estate bond issuing enterprises,which indicates that the per capita income of residents,as the demographic factor and income factor affecting real estate cash flow,can be included in the credit risk evaluation index system;(2)The model construction shows that SMOTEENN algorithm is effective in solving the unbalanced sample problem of SVM model,and can effectively improve the accuracy of credit risk prediction of SVM model;(3)On this basis,the GA optimization algorithm can further improve the prediction accuracy of the model.The accuracy of GA-SMOTEENN-SVM model is 96.15%,the recall rate is 100%,and the F1 value is 98.04%;(4)The empirical research shows that the prediction results based on GA-SMOTEENN-SVM model have significant statistical significance,and can provide an operable application method for preventing and dealing with enterprise credit risk.The research innovation of this paper is as follows:(1)Innovation from the research perspective: When previous scholars studied enterprise credit risk,they used ST and non ST as the criteria to measure whether an enterprise has credit risk.In this paper,the credit risk of real estate enterprises is considered,and the credit risk variables of debt issuing enterprises are the existence of default of credit bonds or the credit rating of debt issuing enterprises is Baa3/BBB-and below.(2)Innovation of research indicators: based on the fact that population factors and income factors are important factors that determine the cash flow,solvency and financing capacity of real estate enterprises,the growth rate of per capita income of residents is included in the credit risk indicator system,which enriches the indicator system.(3)Research model innovation: SMOTEENN sampling algorithm is used to process the unbalanced data,and genetic algorithm is used to optimize the parameters of the traditional model to build the GA-SMOTEENN-SVM model of real estate enterprise credit risk early warning. |
