An Empirical Study On The Combined Early Warning Model Of Financial Crisis In China's Listed Companies

Under the current conditions of the socialist market economy in our country,the competition among enterprises has become increasingly fierce.It is not uncommon for companies to declare bankruptcy due to various reasons.However,the financial crisis is a gradual process of development.Generally speaking,it takes at least a few years for companies to go from normal to worsening and eventually to bankruptcy.It can be seen that corporate financial crisis can be forecasted and prevented in advance.Therefore,in order to enable enterprises to recognize and guard against the financial crisis earlier,so that investors can reduce investment risks,enable the government to accurately assess market risks,ensure the stable development of market economic entities,and establish an efficient financial crisis early warning model is of great significance.This article reviews domestic and international research on financial crisis early warning.Based on the existing research,we selected a sample of 32 special treated companies and 128 matching companies,using logical regression,decision tree,support vector machine model,and combining early warning model based on the above three models to conduct empirical research,and establishing an early warning model with high prediction accuracy.This empirical study is mainly divided into three phases:First,select indicators from five aspects:profitability of listed companies,debt repayment ability,operational ability,development ability,and cash flow,and select 22 primary financial indicators through factor analysis.Index optimization was performed and eight common factors were extracted as input variables for follow-up empirical analysis.Then,based on 2014(T-3)data of samples as valid year data,and use 2015(T-2)years of data as comparison data,and secondly,to overcome The unbalanced problem of the sample set,the use of the smote algorithm to optimize the training set,based on the above indicators and data for the empirical research of logistic regression,decision tree and support vector machine model Studies have shown that in three separate models,the support vector machine model has a higher prediction accuracy and a more stable effect.Finally,using the discriminant results of the above three models as input variables,two kinds of combined early-warning models were established.The results show that the accuracy of the two combined early-warning models has reached more than 85%for the samples selected in this paper.It can be concluded that the effect of the combined early warning model is better than that of the single early warning model,and based on the indicator system and the data of the year of selection in this paper,good prediction results can be achieved.The early warning model established in this paper is effective and practical.According to the company's specific situation,the company's managers can refer to this study to quantify the future financial status of the company,at the same time,combine other non-quantitative factors to analyze the causes of the financial crisis to achieve timely warning and effectively reduce the financial crisis,and protect the interests of investors,companies,and various departments.