The Research On Financial Early-warning Based On Data Mining Algorithm

Chinese A-share listed companies play an important role in the development of national economy and have made important contributions to global capital allocation.However,the listed companies face complex and varied internal and external environments.It is common for A-share companies fall into financial crisis due to improper management decisions.Therefore,it is of great significance for owners,managers and other stakeholders to establish an effective financial early warning system to predict the future financial situation of companies,so as to avoid the financial crisis.This thesis investigates the financial early-warning for A-share listed companies,taking the special treatment of listed companies due to abnormal financial situation as the financial distress sample from 2014 to 2017.Based on theoretical analysis,45 variables are selected from both financial and non-financial aspects,including debt paying ability,profitability,operating capacity,development capacity,cash flow,per share index,governance structure,ownership structure and audit opinions indexes.After the significance test,optimal discretization,and neural network financial early warning model in year(T-3),(T-2)and(T-1)for the 45 variables,the indicators that have significant impact on the financial situation in three years have been selected,which are working capital turnover,net interest rate on total assets,growth rate of operating profit,earnings per share,net cash flow from operating activities per share,and audit opinions.Finally,the validity and briefness of these indicators have been verified by decision tree analysis and logistic regression for financial early warning model.In addition,according to the attribution of the above six indicators,this thesis puts forward suggestions to improve the financial situation of companies from five aspects:operation ability,profitability,development ability,per share index and audit opinions,so as to avoid financial distress.