Forecast Of Guangdong Ocean GDP Based On BP Neural Network

Gross ocean production is a quantitative index of the development and utilization of marine resources, which include all kinds of economic activities and related industries in the national economy, also is the reaction of sea related economic activities for a period of time. At the same time, Gross ocean production is an important indicator of the total marine production and GDP accounting. In 2005, China’s marine GDP was ￥ 1698.7billion, which accounted for 4% of gross domestic product. Ten years later, in 2014, the GOP(Gross ocean production) of China is ￥5993.6 billion, which accounted for 9.4% of GDP, and to maintain the growth rate of approximately 7.6% on increasing. Actually, the total marine production is one of the important indexes which can directly affect the national macroeconomic stability and health. Therefore, on the basis of the total marine production operation rules, forecasting the gross ocean production of marine economy and even the national economy ‘sustained and healthy development will have a important guiding significance.This paper first introduces the research’s background, research significance, research methods and research ideas of this paper. Not only having a comprehensive study of the theory from predecessors and some models of marine GDP forecast gross ocean production, but also informed the significance of forecasting gross ocean production and achieve results. In this paper, we introduced the BP neural network model, which have a stronger fitting ability, and applied it to the actual forecast of Guangdong’s ocean GDP. In 2001 and 2006,the way of ocean economic statistics have been changed in some degree., besides, there are so many date included. Therefore, in the actual process of modeling, not only facing a small number of marine data but also many complex types of characteristics. For solving these problems, in this paper we applied principal component analysis to choose five key factors of Guangdong marine GDP, which have a high degree of influencing, and can basically represent the total information in gross ocean economic production, Finishing that, we solved the character of "complex" in the statistics. Secondly, applying Bagging technology in generating learning samples to train the BP neural network, solved the problem of leaking available statistics. Finally, BP neural network prediction may face the problems of unstable, so the integrated BP neural network is introduced to improve the accuracy of the BP network model in the prediction of the gross ocean economics production. In addition, in order to verify the feasibility and reliability of the prediction model, this paper also will apply the exponential smoothing model to verify that the BP neural network’s predict effect in the gross ocean economic production. Besides, we have a conclusions, which specially refer in prediction of the gross ocean economic production, prediction model based on BP neural network has a great advantage of accuracy compared to traditional exponential smoothing model. But there are still some problems which is difficult to explain the relationship between the variable and other defects. Finally, the author also discussed the chances based on the BP network model in transformation and application.