| To enhance mutual understanding of a region’s economic development level,grasp its economic development situation,and better help decision-makers to make the next regional economic development planning and deployment,it is of self-evident importance to establish a model close to the actual data change trend.Firstly,in the selection of regional economic data,gross regional product(GDP),as a classical economic statistic,represents the market value of all final goods and services produced in a certain period in a region.Meanwhile,In macroeconomics,GDP is also considered to be the best indicator to measure social and economic welfare.Therefore,based on the historical DATA of Suzhou GDP and related economic data,this paper conducts a forecast research on Suzhou GDP.Secondly,in the model selection,this paper takes the grey statistical model as the basic model,and on this basis,expands and improves,as well as the corresponding combined prediction model research.The theoretical system of Grey system began with the article"Control problems of grey systems" published by Professor Deng Julong in Systems and Control Letters in 1982.He creatively proposed grey number and took uncertain information as the research object.Compared with traditional statistical models,grey model has a relatively broad requirement for sample data,so it has a natural advantage in dealing with small sample problems.The annual GDP data of Suzhou and its related factors selected in this paper are taken from the official website of Suzhou Statistics Bureau and the statistical Yearbook of Suzhou in 2020.For the above time series data,the GDP of Suzhou is fitted and predicted through data modeling and programming calculation.The grey GM(1,1)expansion model and background value modification model,grey GM(1,N)modification model,second-order grey difference model,grey-ARIMA coupling model and grey production function model are established respectively.In the grey GM(1,1)expansion model,in order to more accurately express the variation form of high growth of GDP data series through the model form,this paper expands and improves the grey action on the basis of the traditional GM(1,1),and further constructs two expansion models through the different value range of generation coefficient in the background value.In the grey GM(1,1)model based on background value modification,two methods are used to construct the new background value function:on the one hand,conduct function of 1-AGO sequence,the new background value function is obtained by substituting the calculated results into the background value function formula;On the other hand,the method of direct modeling is used to make the original function of the time-response as the background value.In the grey GM(1,N)modified model,constant parameters are added to the right side of the traditional model to correct it,and the first cumulative sequence of the relevant factor series is replaced by its time response equation,so as to improve the prediction accuracy while simplifying calculation.In the second order grey difference model,the difference property of the grey model is used to build the DEDGM model,and the difference equation theory is used to solve the model directly,which can ensure the consistency of the form before and after the equation.In the Grey-Arima coupling model,on the basis of GM(1,1)extended model fitting prediction,the residual column is obtained and analyzed by ARIMA model.Finally,the residual prediction results of ARIMA model are used to modify the grey prediction results,so as to construct a Grey-Arima coupling model better than the single model.In the grey production function model,the grey idea is integrated into the two-level nested CES production function in this paper.There are no grey parameters in the model,but the grey model is preprocessed on the original data,so as to eliminate the interference brought by certain random fluctuations and contribute to more accurate estimation of the subsequent parameters of the model.The new model is used to demonstrate the annual GDP data of Suzhou respectively.The results show that the new model has high simulation accuracy and prediction accuracy,and the error of each improved optimal model is basically less than 3%,which is significantly better than the traditional model.Among them,DEDGM(2,1,a1k,a2k,bk)model has the most ideal effect.Therefore,this model is finally selected to forecast the GDP of Suzhou city in 2021-2025.At the same time,we believe that the prediction model can also provide a strong theoretical guidance for the regulation and policy making of China’s regional economy in the new era. |
