Bayesian decision theoretic scale-adaptive estimation of a log-spectral density

The problem of estimating the log-spectrum of a stationary time series by Bayesian shrinkage of empirical wavelet coefficients is studied. A model in the wavelet domain that accounts for distributional properties of the log-periodogram at levels of fine detail and approximate normality at coarse levels in the wavelet decomposition, is proposed. The smoothing procedure, called BAMS-LP (Bayesian Adaptive Multiscale Shrinker of Log-Periodogram), ensures that the reconstructed log-spectrum is sufficiently noise-free. It is also shown that the resulting Bayes estimators are asymptotically optimal (in the mean-squared error sense). Comparisons with non-wavelet and wavelet-non-Bayesian methods are discussed.

Overview