Research On Financial Crisis Early Warning Of Private Listed Companies Based On Factor Analysis And XGBoost

With the development of the national economy of china and the deepening of institutional reform,the organizational structure of private listed companies is gradually being optimized,the financing scale is also expanding and the financing proportion has increased significantly.They have played a very important role in promoting the healthy development of the national economy.With the rapid and healthy development of market economy,the competition is becoming increasingly fierce.Private listed enterprises are inevitably faced with various challenges and threats.Enterprises can cope with these threats and survive in the competition,which will affect the healthy development of the whole society and economy.Therefore,it is very important to carry out early warning discrimination on the potential threats or potential crises faced by enterprises.Through early warning discrimination,enterprise managers can find the enterprise crisis as early as possible,so as to adjust strategies and take effective measures to prevent or deal with the crisis.Therefore,the establishment of financial crisis early warning model to explore the causes of financial risk and countermeasures has great research significance and practical value.This paper selects 190 private enterprises as research samples,which contains 42 of the financial crisis of enterprises,148 normal financial enterprises.In addition,47 financial indicators cover eight aspects such as enterprise development ability,operation ability and profitability and so on.This paper mainly studies the statistical test of financial indicators,the construction of the index system and the construction of the financial early warning model.(1)Statistical test of indicators and construction of secondary index system.Firstly,the sample data are preprocessed to effectively ensure the accuracy and integrity of data.Secondly,the normality test and significance test of the financial indicators are carried out by using the statistical test method.According to the test results can be seen,with 28 indicators in 47 financial indicators passed the test of significance.Then factor analysis is used to further reduce the dimension of the 28 financial indicators that have passed the test,and 9 common factors are extracted to form the second-level financial indicator system.(2)The construction of financial early warning model.According to the factor scores of 190 private listed companies on 9 new variables,taking the variance contribution rate of factors as the weight of 9 new variables,the comprehensive scores of 190 samples are calculated,and the financial early warning discriminant model based on the factor scores is established.The results show that the prediction accuracy of this model is 79.47%.Then,based on the above index system,taking the nine factors as the new variables,70% of the190 research samples are divided into training set and 30% were divided into test set.At the same time,combined with the confusion matrix,ROC curve and AUC values to discriminant XGBoost and Logistic regression model,then the prediction accuracy of the financial early warning models is compared and analyzed.The results show that the financial early warning model based on XGBoost is more accurate.Finally,the research limitations of this paper are pointed out,and the future research prospects are put forward.