Study On Construction And Application Of The GDP Combination Forecasting Model

Gross domestic product(GDP)is the core indicator of the national economic calculation system.GDP can accurately measure the macroeconomic development of a country,and it can provide a strong theoretical support for the country to formulate correct economic development strategies and policies.In recent years,the situation of international economic recovered slowly,and China's economy is currently undergoing three phases of superposition and is affected by a series of structural adjustments.China's economic situation has generally shown a downward trend of development,economic development has entered the “new normal” and the GDP growth rate has continued to decline slightly.The “13th Five-Year Plan” period is a crucial five-year period for China to constructing a all-round well-off society.Therefore,under the new economic situation,this paper has important guiding significance for realizing the high-precision prediction of GDP in the “13th Five Year Plan” period in China.In order to achieve high-precision prediction of China's GDP,this paper puts forward a new combination forecasting model based on model screening and Markov chain optimization IOWA operator.First,the article selects six single models for GDP forecasting,namely ARIMA model,double exponential smoothing model,GM(1,1)model,gray Verhulst model,GM(1,1)power model,and BP neural network.Secondly,a comprehensive validity index is constructed to screening the single model based on the time-weighted grey correlation degree index and the forecast validity index,and the redundant check is used to screening the model further.Then,the single model selected is used to establish four combination forecasting models based on variance-covariance,entropy,correlation coefficient and IOWA operator.The combination model based on the IOWA operator can orderly assign the weight to the single model.At this time,the weights and the model are no longer corresponding to a fixed relationship.However,the prediction accuracy of the model during the prediction period cannot be known,it is impossible to determine the order of the orderly weighting of the model in the forecast period.The traditional solution is to arithmetically average the historical weight coefficient and to use it as the weight coefficient in the future,but this method of processing lacks sufficient theoretical basis.Finally,aiming at this deficiency,this paper uses Markov chain to optimize the IOWA operator and proposes a combined forecasting model based on Markov chain optimization IOWA operator.Through analysis of China's GDP forecasting,and comparing the forecasting effect of each model under the established evaluation index system.The results show that the new combination forecasting model established in this paper has higher prediction accuracy than the six single model and the four combined models.It can significantly improve the forecasting effect of the combination model based on traditional IOWA operator.According to the new model prediction,the total GDP of China from 2018 to 2020 is 904182.3,982346.7,and 1085541.2 billion yuan respectively.