| One cannot stand oneself without trustworthiness,and an industry cannot prosper without commercial credit.Since the 18 th CPC National Congress,the government make much account of credit construction.Credit is the root of the market,the foundation of society and the foundation of China standing in the jungle of the world economy.Behind China’s business form is accompanied by Chinese culture,which has always stressed the supremacy of integrity.For enterprises,the code of credit ethics is ought and necessary.It can bring invisible economic benefits to enterprises.To some extent,credit resources are more important than human and material resources.Listed companies,as the "bellwether" of China’s commercial system,have been playing an indispensable role.After the occurrence of COVID-19,the credit risk events of listed companies in China are frequent,which hinders the stable development of the financial system to a certain extent.Therefore,this article selects all non ST companies in the CSI 300 index and all ST enterprises in the listed companies from the Wind database as the research object,takes all the financial information of these 490 enterprises from the first quarter of 2019 to the first quarter of 2021 as the data set,studies the credit risk of listed companies combined with Deep Belief Network and Structural Equation Model,establishes an forewarning mechanism and puts forward preventive suggestions.The main content of the paper includes the following three parts:A credit risk forewarning model based on Deep Belief Network is constructed.The complex credit data of listed companies makes it difficult to study their credit risk.With the help of the Deep Belief Network method,optimize various parameters in the network structure to train the Deep Belief Network,and learn the credit risk forewarning network model of listed companies with three hidden layers.A credit risk forewarning model based on DBN-SEM is constructed.Through network model learning,the top 20 neurons of network weight parameters in the input layer are inversely found,and their corresponding indicators are used as credit risk forewarning indicators to summarize the forewarning indicator system that has an important impact on credit risk.Reveal the internal structure of these 20 indicators and find the common factors behind them,according to the factor analysis.Takes the extracted three common factors as the latent variables of the structural equation model,constructs the DBN-SEM based credit risk forewarning model of listed companies with three latent variables and seven explicit variables,finds out the hidden variables behind the forewarning indicators,measures the important forewarning indicators of credit risk,and analyzes the influence path of credit risk.According to the above experimental results,forewarning and prevention are put forward.Analyze the results of model learning,and obtain the forewarning value through its changes with the help of the constructed key forewarning index system of credit risk.Through the learn of the deep meaning of various index of the double model,the relationship between potential variables is extracted,and targeted credit risk prevention suggestions are put forward. |
