A0170
Title: Bayesian nonparametric spectral analysis of locally stationary time series
Authors: Renate Meyer - University of Auckland (New Zealand) [presenting]
Yifu Tang - University of Otago (New Zealand)
Claudia Kirch - Otto-von-Guericke University Magdeburg (Germany)
Kate Lee - The University of Auckland (New Zealand)
Abstract: Based on a novel dynamic Whittle likelihood approximation for locally stationary processes, a Bayesian nonparametric approach to estimating the time-varying spectral density is proposed. This dynamic frequency-domain-based likelihood approximation characterizes the time-frequency evolution of the process by utilizing moving periodograms previously introduced in the bootstrap literature. The posterior distribution is obtained by updating a bivariate extension of the Bernstein-Dirichlet process prior with the dynamic Whittle likelihood. Asymptotic properties such as sup-norm posterior consistency and L2-norm posterior contraction rates are presented. Additionally, this methodology enables model selection between stationarity and non-stationarity based on the Bayes factor. The finite-sample performance of the method is investigated in simulation studies, and applications to real-life data sets are presented.
