The Empiricical Research And Prediction Of CPI Based On The Wavelet Analysis

In macroeconomic economic theory, price stability is one of the four economic objectives. The consumer price index, the CPI, is a measure of the main indicators of the price level is an important indicator of economic theory. CPI index has been used widely, the consumer price index can be observed and analyzed the impact of change on money wages, annual rates of CPI change are usually calculated as a measure of inflation indicators, so that the world is usually measured by CPI inflation the degree of swelling. CPI trend not only to macroeconomic policy-making and other countries into account, but also to residents of clothing, housing and other livelihood. CPI vane both national macroeconomic trends, but also the cost of changing people's lives compass.First, the study of the relevant CPI forecast a more systematic literature review, the contribution of previous learning and research. Then the theory to be adopted on this article, methods and models are introduced in detail. Finally, select January 2001 to June 2010 the CPI data for empirical analysis. After analysis, we use time series analysis methods of uncertainty representation model of Holt-Winters model and the stochastic analysis of ARMA time series modeling of the CPI model to predict, respectively, and then the CPI to 3 layers wavelet decomposition, the decomposed approximation and details first with Holt-Winters model fitting and prediction, and then reconstruct the predicted CPI; Similarly, after the use of ARMA (GARCH) model and approximate the decomposed parts of the details of fitting and forecasting, re-structure predicted by CPI. By comparing these four predictive models and concluded:Whether long-term projections or short-term forecasts, wavelet 1 model (cited as the Holt-Winters model wavelet) the prediction effect is the best; wavelet 2 model (ARMA introduced wavelet (GARCH) model) and Holt-Winters forecasting model is better in the short term, and long-term forecasting ARMA model is better.