Application Research Of Time Series Analysis In The CPI

With the rapid development of science and technology, systems theory constantly developed and improved. Mathematical modeling of complex systems attracted more attention, in particular, the establishment of complex mathematical models systems that brought uncertain elements requires the use of probabilistic statistical models. Time series analysis plays an important role in the modeling and analysis as a powerful tool, and particularly has been widely used in the field of financial data analysis and signal processingIn the analysis of financial data, the consumer price index CPI is an important measure of the price level and indicator of economic theory. The CPI can measure the extent of inflation in the country, so the CPI is not only related to the formulation of macroeconomic policies, but also to clothing, housing and other livelihood of residents. Based on the CPI research in the academia, people mainly used in the classical time series analysis, nonparametric estimation methods and new data mining methods.No matter what the time series is deterministic or stochastic time series methods by the study of the classical time series analysis methods, we found that it based on certain basic assumptions, such as stability and normality etc. while the CPI data usually dose not meet these assumptions, the main problem is concentrated in the pre-processing of data. Regardless of the classical decomposition of time series analysis how to disaggregated data, It cannot accurately predicated the result. This article gets to the prediction accuracy close model on the time series processing by analysis of s data, data multi-scale decomposition and reconstruction, combined with classical time series forecasting model, and compared the fore-and-aft introduction of wavelet analysis.Based on the above idea, first the article minute introduced the classical time series model. second it introduced the theory of the origin of the Fourier transforming and wavelet analysis. Finally, it calculated modulus value as CPI forecast value by the CPI data analysis of Jilin Province, and random selected ARIMA models for time series analysis and forecasted modeling CPI in the classic model, and then respectively analyzed the Fourier transform and wavelet analysis by the Fourier transform of the data. As follow the real and imaginary parts of the classical time series model respectively analyzed and predicted future values and performed an inverse Fourier transform. Wavelet analysis predicated value and reconstruct the CPI by one times wavelet decomposition, fitting and forecasting the decomposition of the approximation and details By compared the effect of various model predictions and drawed conclusions, the introduction of wavelet analysis that is better model predicated result...