A1054
Title: Non-parametric mixture models for covariance function estimation
Authors: Stephen Berg - Penn State University (United States) [presenting]
Abstract: An approach for estimating covariance functions based on nonparametric mixture models will be introduced, with an emphasis on a weighted least squares estimator of the autocovariance sequence from a reversible Markov chain. The estimator is shown to lead to strongly consistent estimates of the asymptotic variance of the sample mean from an MCMC sample, as well as to consistent estimates of the autocovariance sequence. An algorithm for computing the estimator is presented, and some empirical applications will be shown. The proposed shape-constrained estimator exploits a mixture representation of the autocovariance sequence from a reversible Markov chain. Similar mixture representations exist for stationary covariance functions in spatial statistics, including for the Matern covariance as a special case, and some extensions of shape-constrained approaches are highlighted for estimating covariance functions in spatial statistics.
