Risk Prediction Method of Village Banks Based on Equity Relationship

In the context of the simultaneous occurrence of financial risks in four rural banks in Henan, China, the public is worried about rural banks. Unlike big banks, which are “too big to fail”, rural banks face the problem of “too many to fail”. In this paper, an indicator system is established by selecting all the beneficial shareholders and their corresponding share data in the corporate equity of 82 banks. Based on the decision tree, KNN algorithm and RUSBoost algorithm, a risk early warning model for Henan rural banks is constructed, and it is verified that the three algorithms can be used to a certain extent. Two states of risk and normal were classified, and the RUSBoost model performed the best, with a false positive rate (FPR) value of 7% and a false positive rate (FPR) value of 0. The control group was introduced, and the equity data that was not processed by the index system was studied based on the same algorithm, and it was concluded that the index system had a good effect. This is the first attempt to identify the risk status of rural banks in my country by building an equity structure index system based on machine learning technology and will provide a prototype technical means and thinking direction for rural bank operators and regulatory authorities to provide timely early warning.