Factors and Impacts of Financial Crisis Formation Based on Big Data and RF Early Warning Model

Abstract

With the continuous changes and globalization of financial markets, the factors that shape financial crises have become more complex. To solve the unbalanced data, the study introduces the synthetic minority over-sampling technique with editorial nearest neighbour method to improve the performance of the model. By examining the financial indicators in detail, different aspects of ST and non-ST enterprises are compared. The results show that the synthetic minority over-sampling technique and the editorial nearest neighbour method perform well in financial risk prediction, improving the correct classification rate of the "1" sample by 2.26% and the "0" sample by 3.46%. In addition, the study observes that the "cash content of operating income" of special treatment enterprises is significantly higher than that of non-special treatment enterprises, and that there are significant differences in debt service capacity, overhead growth rate, and profitability. The study provides a powerful methodology for corporate financial risk management and a more effective risk monitoring tool for government regulators and financial institutions.