Influence Factors And Prediction For The Revenue Of Yunnan Province Based On Variable Selection And Neural Network Methods

Local fiscal revenue can fully reflect the operation status of the regional national economy,and it is also an important issue for the government’s macro-control in the market economy.With the initial establishment of a socialist market economy with Chinese characteristics,especially since China entered a new era,the analysis and forecast of local fiscal revenues has received increasing attention from all walks of life.The research results on the analysis and forecast of fiscal revenue at the national level are numerous,but most of the articles use a single model to analyze and forecast the fiscal revenue,and there are few articles that specifically focus on the analysis and prediction of Yunnan Province’s fiscal revenue impact variables.This paper first establishes a multivariate linear regression model of fiscal revenue(y)and 10 related independent variables in Yunnan Province,and uses the lasso method to perform variable selection on multiple linear regression models.Since the data itself is time-series data,a single grey prediction model GM(1,1)is established for the Yunnan provincial fiscal revenue data,and the grey system GM(1,1)and BP neural network are established based on the selected key variables.Combined forecasting model.Using the historical data of the fiscal revenue of Yunnan Province to empirically analyze the two established models,and comparing the prediction errors of the two models,the model with a smaller prediction error is the final Yunnan Province fiscal revenue forecast model,and based on the final model,it is given 2017-2019 Yunnan Province fiscal revenue forecast value.