The Study On Applying Random Forest To Forecasting The Financial Distress Of Listed Company

When a listed company runs into the situation where its cash flow is less than the current debt,the relative value of the its debt is negligible,or its current assets are less than the current liabilities,the company is said to be in a financial distress.The financial distress will put the company's stakeholders including investors,creditors and suppliers at risk.With the progress of removing production overcapacity and restructuring the macro economy in today's China,the chance for a list company to run into financial distress has been increasing considerably.As such,for financial or investment institutions as well as creditors,it is imperative to accurately forecast a list company's financial distress.Meanwhile,the recently emerging online financial platforms are also in demand of reliable forecast to thereafter work out effective countermeasures.This paper is dedicated to forecasting the financial distress of listed company through building and applying random forest model.The proper definition of financial distress is the foundation to build a financial distress forecast model.This paper innovatively defines the financial distress by studying the listed Chinese manufacturing companies.On top of the current definition that only takes ST companies as samples,or only employs indicators that are inadequate in reflecting liquidity,the new definition that is raised in this paper also takes the list company's cash flow,dividend and profit into consideration.The paper not only studies the traditional financial indicators but also works out more non-traditional financial indicators through the feature engineering(to extract features from the raw data to the best extent for algorithm and model use)for model construction by means of relative models and methods.It is found out that these non-traditional indicators are more effective and fruitful in the case of financial distress forecast.The study reveals that random forest model is more effective in forecasting the financial distress of listed company accurately,compared with traditional logistic model.In conclusion,the random forecast model has a better performance,and shall be adopted more in forecasting a list company's financial distress.