Journal of the Korean Statistical Society

The Korean Statistical Society (한국통계학회)
권호 표지

An Automatic Spectral Density Estimate

  • Park, Byeong U. (Department of Computer Science and Statistics, Seoul National University, Seoul 151-742) ;
  • Cho, Sin-Sup (Department of Computer Science and Statistics, Seoul National University, Seoul 151-742) ;
  • Kee H. Kang (Department of Computer Science and Statistics, Seoul National University, Seoul 151-742)
  • Published : 1994.06.01
Download PDF
⟨ Previous Next ⟩

Abstract

This paper concerns the problem of estimating the spectral density function in the analysis of stationary time series data. A kernel type estimate is considered, which entails choice of bandwidth. A data-driven bandwidth choice is proposed, and it is obtained by plugging some suitable estimates into the unknown parts of a theoretically optimal choice. A theoretical justification is give for this choice in terms of how far it is from the theoretical optimum. Furthermore, an empirical investigation is done. It shows that the data-driven choice yields a reliable spectrum estimate.

Keywords

References

  1. Journal of Time Series Analysis v.8 Determining the bandwidth of a kernel spectrum estimate Beltr$\~{a}$o,K.I.;Bloomfield,P.
  2. Time Series Data Analysis and Theory Brillinger,D.R.
  3. Journal of American Statistics Association v.80 Data-driven choice of a spectrum estimate:Extending the applicability of cross-validation methods Hurvich,C.M.
  4. On estimating integrated squared spectral density derivatives Lee,Y.H.;Cho,S.;Kim,W.C.;Park,B.U.
  5. Spectral Analysis and Time Series Priestly,M.B.
  6. Econometrica v.59 Automatic frequency domain inference on semi-parametric and nonparametric models Robinson,P.M.