In order to find the most suitable model for regional economic prediction, a regional economic development prediction based on support vector machine and random forest algorithm is proposed. By comparing the fitting accuracy of each model, the mathematical model that is most suitable for the economic forecast of a province is found, and the GDP of a province is predicted. The results show that the predicted value of multiple linear regression is quite different from the real value. The main reason is that the GDP growth of a province has been extremely slow since 2019, and even negative growth has occurred in 2021. Therefore, the overall data shows nonlinear characteristics. While multiple linear regression analysis is mainly used to study linear models, the effect is not very ideal for nonlinear models. It is proved that this method has important reference value and great guiding role for the local government to regulate and control the local economy.