Impact Analysis of Financial Early Warning Indicators Based on Random Forest

Abstract

In order to improve the indicator selection method for financial early warning, this paper combines the idea of K-fold cross-validation to improve the sampling method of Random Forest (RF) and proposes the K-fold random forest algorithm (KRF). The experimental results show that the KRF algorithm has a better classification performance than the RF algorithm, and improves the accuracy of the RF algorithm on the indicator. Finally, the importance of the selected financial indicators to the financial early warning is determined. A more scientific and accurate indicator system will provide a research basis for further financial early warning research.