A Study on the Incremental Size of Social Financing Based on XGBoost and SHAP

This study applies machine learning methods to predict the scale of social financing, which is an important basis for formulating monetary policy and conducting macroeconomic regulation. The results show that XGBoost has high accuracy and stability in predicting the incremental size of social financing, and the SHAP values help to explain the main factors influencing the model predictions. The results show that in addition to RMB loans, important incremental features of the social financing scale include trust loans, corporate bonds, and undiscounted bank acceptance bills. This study provides a feasible forecasting method that can help governments, enterprises and investors formulate appropriate economic policies and measures, optimise business strategies and select appropriate investment strategies.