Construction And Empirical Study Of GDP Forecasting Model Based On Geographical Space

Gross national product(GDP)is a basic index to measure the state of the country's economy.Its stable and sustained growth is closely related to the healthy development of the national economy.The GDP forecast is related to that the government distinguish whether the economic state is shrinking or expanding.The final decision is whether to stimulate or restrain economic growth,which plays an important role in the overall macroeconomic research.However,GDP data has the characteristics of time series and non-linearity,and its forecasting is relatively complicated.There are two main problems in the existing research: First,the GDP forecasting accuracy is lacking;second,the model is only applicable to specific regions and can not complete regional economic forecasting according to user expectations.In view of the above problems,based on previous studies,this paper combines spatial econometrics with artificial intelligence in order to improve the accuracy of macroeconomic forecasting.In the selection of indicators,the geospatial indicators are introduced into GDP forecasting and a set of scientific and reasonable GDP forecasting indicators system is constructed to eliminate the impact of regional differences on economic forecasting and achieve the GDP forecasting in the regions expected by users.On the forecasting model,based on BP neural network,the forecasting accuracy of the model is improved by double optimization.Firstly,according to GDP accounting method and related factors,this paper identifies the basic indicators of GDP forecasting,and introduces geographical indicators into GDP forecasting by numerical method,and establishes a scientific index system of GDP forecasting based on geographical space.Then,on the basis of BP neural network model,this paper establishes a double optimization BP neural network model.The model is mainly divided into two submodules: firstly uses the dynamically adjusted learning factor to optimize the particle swarm optimization algorithm,and then uses the optimized global optimization ability of the optimized PSO algorithm to continue to optimize the weight and threshold of the BP neural network.In the BP neural network module optimized by particle swarm optimization,the optimal position of the population obtained by particle swarm optimization is mapped to BP neural network,replacing the original randomly selected weights w and biased b,thus reducing the training burden of the neural network and improving the forecasting accuracy of the model.At the same time,the learning factors c1 and c2 are adjusted dynamically to accelerate the convergence speed of the particle swarm optimization algorithm and complete the optimization of the whole model.Finally,through the empirical analysis of two important indicators: GDP growth rate and per capita GDP,it is found that the proposed GDP forecasting model based on geographical space can indeed improve the accuracy of GDP forecasting,and at the same time,it can achieve the GDP forecasting of users' expectations.It is concluded that China should increase investment in education,science and technology as well as talent introduction in the remote areas of the north and west,and stimulate the economic growth of underdeveloped areas by stimulating residents' consumption and expanding import and export.