GDP Forecast Of Jilin Province Based On Bpneural Network Model

Gross domestic product not only is the focus factor, of which is about the contemporary national economic accounting system, but also is the primary factor of weighing a country of the comprehensive national strength. This factor puts all the results of the national economy output in a refine statistics number. It also one of provides the most comprehensive standards to assess and weigh, the national economic development status, the trend of GDP growth and social wealth display. It is, as it were, the most important economic factor which can affect the economic life and social life. So, if a country’s economic polices will be reasonable planning, it is especially important to analyze and research the GDP. It is essential to carry on a model which can forecast reasonably and accurately the future development of the economy. Establish a good economic forecasting model, can accurately grasp the situation of economic development, and set up the corresponding control measures according to the trend, such as the fiscal policy, the regional construction policy, the monetary policy, and so on. In the existing ways of economic forecasting, time series analysis and the regression model are more commonly used. But with the rapid development of economy, the additional interference factors of affecting GDP are increasing, and the change of GDP is nonlinear. So in the time quantum of predicting, the uncertainty of sudden situations take place frequently, it becomes a problem if we use the traditional prediction method of relatively accurate projections for GDP.Artificial neural network is a complex network which is nonlinear, non-locality, the steadi-ness nondeterministic, with parallel information processing structure and the adaptive form of brain information processing ability. So it can charge a large number of common sense by "self learning" or "training" to deal with the specific things. Even more, BP network has obvious superiority in creating forecasting model. On the one hand, it can predict on the basis of the existing data model, and the process is not complicated; On the other hand, it also can close to the useful data intiatively, and find the rule in the data. By experiments, it shows that neural network model is better than the traditional prediction method and is more convenient in the aspect of time series data to predict, especially complicated nonlinear time sequence forecast.Economic forecasting problem is a typical multiple indicators and complex system of small sample prediction problem. In this paper, we preprocess the original data and the stepwise regression analysis with SAS. So it is convenient to study the data. And there are five significant effect factors, includingX2(The total population), X4(Government consumption), X5(visitor arrivals), X6(The tax), X7(The total amount of import and export).According to GNP statistical data of Jilin province during 1992-2011, we can study the related factors which influence the Jilin province economic development situation by using simulation fitting, Neural network prediction or other methods after preprocessing. In this way, we can solve the problem which is Network simulations to predict complex economic problems with too many variables leading to low efficiency. Examples prove that BP is more better than the traditional Economic Forecast Model in Multifarious economic system simulation prediction. In this paper, it is outstanding to preprocess variables with stepwise regression method, so it can reduce the input variable on the premise of guarantee of information.