Spectral Analysis and Filtering

  • Robert H. Shumway
  • David S. Stoffer
Part of the Springer Texts in Statistics book series (STS)


The notion that a time series exhibits repetitive or regular behavior over time is of fundamental importance because it distinguishes time series analysis from classical statistics which assumes complete independence over time. We have seen in Chapters 1 and 2 how dependence over time can be introduced through models that describe in detail the way certain empirical data behaves, even to the extent of producing forecasts based on the models. It is natural that models based on predicting the present as a regression on the past such as are provided by the celebrated ARIMA or state-space forms, will be attractive to statisticians, who are trained to view nature in terms of linear models. In fact, the difference equations used to represent these kinds of models are simply the discrete versions of linear differential equations that may, in some instances, provide the ideal physical model for a certain phenomenon. An alternate version of the way nature behaves exists, however, based on a decomposition of an empirical series into its regular components.


Discrete Fourier Transform Frequency Response Function Periodic Component Multivariate Normal Distribution Linear Filter 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer Science+Business Media New York 2000

Authors and Affiliations

  • Robert H. Shumway
    • 1
  • David S. Stoffer
    • 2
  1. 1.Division of StatisticsUniversity of California, DavisDavisUSA
  2. 2.Department of StatisticsUniversity of PittsburghPittsburgUSA

Personalised recommendations