Research On Enterprise Financial Risk Early Warning Basedon XGBoost Algorithm

In recent years,the world economic situation is complex.Republican election wins in US presidential election in 2016,Tense situation on the peninsula in 2017,and the rise of trade protectionism in 2018.With the changing environment,listed companies are facing more serious financial risks.It is imperative to give more frequent and accurate financial risk warning to listed companies.At present,many methods and tools have been used in the field of financial risk early warning,such as single factor model,multivariate model,probability model,capital market pricing model,etc.These methods can predict the future financial risk of listed companies to a certain extent,but with the acceleration of environmental change,all aspects will be.Affect the production and operation of enterprises and then affect the financial stability of the company,which requires more effective and adapt to the current environment of financial risk early warning methods and tools.Based on the previous research results and the principles of selecting early warning indicators of financial risk,this paper screens out 107 financial indicators from six dimensions such as profitability and establishes index database.Based on the new XGBoost algorithm,12 important features with more than 5 times of use are selected as independent variables.The characteristic independent variables are analyzed by the optimal box-dividing method,and the XGBoost financial risk early warning model is constructed after repeated learning and training of 2155 stock financial data in Shenzhen and Shanghai Stock Exchanges.In the learning and training process of building the model,the number of control decision trees(42),the depth of the tree(3)and the minimum weight of leaf nodes(1)are selected to optimize the XGBoost model according to the fitting effect of the model.Finally,we choose ST as the case enterprise,and use the optimized XGBoost financial risk early warning model to forecast the financial data of ST Zhongji in 2014,and successfully predict the financial risk,which is consistent with the fact that ST Zhongji sustained losses in 2015 and 2016 triggered financial crisis.This proves that XGBoost model can be used to predict the financial risk of listed companies.At the end of the paper,the author puts forward how to use this model to predict the financial risk of the company and how to deal with the measures after the company predicts the financial risk,and gives some prospects for the shortcomings of the model and the further development.