Study On Financial Crisis Early Warning Of New Energy Industry

As a high-tech industry,the new energy vehicle industry is facing challenges that companies are gradually emerging,and the possibility of financial crisis is gradually increasing.Once the listed company is “ST”,it will aggravate the company’s financial problems again,resulting in a series of vicious circles such as stock price turbulence and financing difficulties,which not only affect the company itself but also bring huge losses to investors,and even affect the financial system,so it is of great significance to establish a systematic and efficient financial crisis early warning model for the new energy vehicle industry.Random forest is a machine learning method developed on the basis of statistics,which has the advantages of not easy to overfit and no strict assumptions,which can solve the problem of early warning of financial crisis.Through the development overview and financial statement analysis of the new energy vehicle industry,combined with the internal,external and transmission risks that the industry may face,this paper believes that the new energy vehicle industry is facing a greater financial crisis.First,the data of 515 companies in the new energy vehicle industry from 2011 to 2021 were selected,including 41 “ST” companies.According to the characteristics of the industry,66 early warning indicators covering the company’s profitability,solvency,operation ability,development ability and R&D ability were selected,and two sets of early warning index systems were screened by the index importance method and the significance correlation method.Secondly,a random forest early warning model is established,the grid search method is used to determine the appropriate parameters,and the two sets of index systems are brought into the model respectively to obtain the model evaluation results such as confusion matrix and AUC.In particular,for the unbalanced data of financial crisis,this paper uses SMOTE technology to process the data.Finally,in order to test the scheme,logistic regression and support vector machine models are added to compare with the random forest model.Through the implementation plan,the following conclusions can be drawn:(1)In the two sets of indicator systems,the indicators representing profitability account for the majority,indicating that the profitability indicators are more representative of the company’s financial status.At the same time,indicators such as earnings per share,interest protection multiple and R&D expense ratio are selected,and the company can pay more attention to such indicators in its operations.(2)In the random forest model,the model early warning effect is better under the index importance method index system,with an accuracy rate of 99.50%.It is explained that in the process of screening indicators,not only the rationality of the indicator system itself,but also the adaptability of the model should also be considered.(3)Compared with logistic regression and support vector machine models,all early warning indicators of random forest have obvious advantages.It shows that this paper uses random forest as early warning of financial crisis in the new energy vehicle industry,which has better early warning results and practical value.