The Application of the Integrated Machine Learning Model in the Financial Crisis of Imbalanced Samples

Abstract In this paper, the financial crisis of the enterprise in the t-period is predicted by the financial index data of the non-financial industry A-share enterprise in the main board market of the Shanghai Stock Exchange in the t-3 period. Then we use the integrated machine learning models to select the financial crisis enterprises, and the problem of failure of classifiers in unbalanced samples is solved through bagging and sampling techniques. The highest probability of correctly selecting enterprises with financial crisis is 92.86%. With the integration of the model, the overall accuracy is improved to 88%. We help the enterprise to complete the forward-looking financial crisis forecast and provide some reference for the business activities and investment activities of the enterprise.