Autoregressive spectral analysis when observations are missing

A new missing data algorithm ARFIL gives good results in spectral estimation. The log likelihood of a multivariate Gaussian random variable can always be written as a sum of conditional log likelihoods. For a complete set of autoregressive AR(p) data the best predictor in the likelihood requires only p previous observations. If observations are missing, the best AR predictor in the likelihood will in general include all previous observations. Using only those observations that fall within a finite time interval will approximate this likelihood. The resulting non-linear estimation algorithm requires no user provided starting values. In various simulations, the spectral accuracy of robust maximum likelihood methods was much better than the accuracy of other spectral estimates for randomly missing data.

[1]  N. Lomb.  Least-squares frequency analysis of unequally spaced data , 1976 .

[2]  J. Scargle.  Studies in astronomical time series analysis. III - Fourier transforms, autocorrelation functions, and cross-correlation functions of unevenly spaced data , 1989 .

[3]  Sina Mirsaidi,et al.  LMS-like AR modeling in the case of missing observations , 1997, IEEE Trans. Signal Process..

[4]  S.M. Kay,et al.  Spectrum analysis—A modern perspective , 1981, Proceedings of the IEEE.

[5]  R. Bos,et al.  AR spectral estimation by application of the Burg algorithm to irregularly sampled data , 2001, IMTC 2001. Proceedings of the 18th IEEE Instrumentation and Measurement Technology Conference. Rediscovering Measurement in the Age of Informatics (Cat. No.01CH 37188).

[6]  Yonina Rosen,et al.  Optimal ARMA parameter estimation based on the sample covariances for data with missing observations , 1989, IEEE Trans. Inf. Theory.

[7]  Piet M. T. Broersen,et al.  Generating data with prescribed power spectral density , 2003, IEEE Trans. Instrum. Meas..

[8]  D. B. Preston.  Spectral Analysis and Time Series , 1983 .

[9]  J. Jouzel,et al.  Climate and atmospheric history of the past 420,000 years from the Vostok ice core, Antarctica , 1999, Nature.

[10]  J. Schoukens,et al.  Frequency domain system identification with missing data , 2000, IEEE Trans. Autom. Control..

[11]  S. de Waele,et al.  Reliable LDA-spectra by resampling and ARMA-modeling , 1998, IMTC/98 Conference Proceedings. IEEE Instrumentation and Measurement Technology Conference. Where Instrumentation is Going (Cat. No.98CH36222).

[12]  Richard H. Jones,et al.  Maximum Likelihood Fitting of ARMA Models to Time Series With Missing Observations , 1980 .

[13]  Lennart Ljung,et al.  An iterative method for identification of ARX models from incomplete data , 2000, Proceedings of the 39th IEEE Conference on Decision and Control (Cat. No.00CH37187).

[14]  Alf J. Isaksson,et al.  Multiple Optima in Identification of ARX Models Subject to Missing Data , 2002, EURASIP J. Adv. Signal Process..

[15]  Piet M. T. Broersen,et al.  The quality of models for ARMA processes , 1998, IEEE Trans. Signal Process..

[16]  D. Rubin,et al.  Statistical Analysis with Missing Data. , 1989 .

[17]  Alf Isaksson,et al.  Identification of ARX-models subject to missing data , 1993, IEEE Trans. Autom. Control..

[18]  M. Loutre,et al.  Spectral analysis of climate data , 1996 .

[19]  Piet M. T. Broersen,et al.  Error measures for resampled irregular data , 2000, IEEE Trans. Instrum. Meas..

[20]  Lennart Ljung,et al.  System Identification: Theory for the User , 1987 .

[21]  Piet M. T. Broersen,et al.  Autoregressive spectral estimation by application of the Burg algorithm to irregularly sampled data , 2002, IEEE Trans. Instrum. Meas..

[22]  Piet M. T. Broersen,et al.  Automatic spectral analysis with time series models , 2002, IEEE Trans. Instrum. Meas..

[23]  J. P. Burg,et al.  Maximum entropy spectral analysis. , 1967 .

[24]  Piet M.T. Broersen,et al.  Spectral analysis of irregularly sampled data with autoregressive models , 2002 .

[25]  B. Anderson,et al.  Optimal Filtering , 1979, IEEE Transactions on Systems, Man, and Cybernetics.

[26]  Piet M. T. Broersen,et al.  The Burg algorithm for segments , 2000, IEEE Trans. Signal Process..

[27]  Richard A. Davis,et al.  Time Series: Theory and Methods , 2013 .