A Study On Application Of Artificial Neural Network For CPI Forecast

Economic growth,full employment,consumer price stability and balance of Payments are the four targets of macro-control.The theorem confirmed and experience studying about price inflation is the problem concerned by academics and government. From the reform and opening up,China has experienced four serious inflations and one a little serious deflation.In the 2007 year,the consumer price had a significant rising .The CPI had rose 3.2% in the same period last year in the first half year.The rising range increased 1.9% than the same period last year.The government should take some effective macro-control measures timely.But,before this ,it must have a prediction for the economics trend .when we have a good prediction,the government's macro-control measures can play a role fully.So,it is significant to study the consume price and prediction technology .Studying the inflation through the quantitative model has become the mainstream method in the modern world.Traditional prediction technologies are most concentrated in the analysis about regression model of causal relationship and time series model.CPI is macroeconomic and nonlinear system ,which has the problem of multiple collinearity and error series related in the actual process of applying.so it is inevitable that lost the information ,and has a bad simulation results.Artificial Neural Network is very effective for establishing prediction model.It can study previous experience automatically from sample data and needn't complex enquiries or formulation process.And it can approximate those best rules implicated by the data automatically.The Artificial Neural Network is better than traditional prediction technologies in the prediction ,especially,in the very complicated nonlinear area.In this paper,we will construction CPI prediction model by the Artificial Neural Network,and use the CPI data of Guangdong provence to do a empirical analysis.This can get a good prediction model ,which will be compared to traditional prediction technologies .That we can see Artificial Neural Network has a good accuracy.