Sequential Empirical Bayes Method for Filtering Dynamic Spatiotemporal Processes

Evangelos Evangelou, Vasileios Maroulas

Research output: Contribution to journalArticle

2 Citations (Scopus)
100 Downloads (Pure)

Abstract

We consider online prediction of a latent dynamic spatiotemporal process
and estimation of the associated model parameters based on noisy data.
The problem is motivated by the analysis of spatial data arriving in
real-time and the current parameter estimates and predictions are updated
using the new data at a fixed computational cost. Estimation and
prediction is performed within an empirical Bayes framework with the aid
of Markov chain Monte Carlo samples. Samples for the latent spatial field
are generated using a sampling importance resampling algorithm with a
skewed-normal proposal and for the temporal parameters using Gibbs
sampling with their full conditionals written in terms of sufficient
quantities which are updated online. The spatial range parameter is
estimated by a novel online implementation of an empirical Bayes method,
called herein \emph{sequential empirical Bayes} method. A simulation
study shows that our method gives similar results as an offline Bayesian
method. We also find that the skewed-normal proposal improves over the
traditional Gaussian proposal. The application of our method is
demonstrated for online monitoring of radiation after the Fukushima
nuclear accident.
Original languageEnglish
Pages (from-to)114-129
JournalSpatial Statistics
Volume21
Issue numberA
Early online date29 Jun 2017
DOIs
Publication statusPublished - 1 Aug 2017

Fingerprint

Spatio-temporal Process
Empirical Bayes Method
Dynamic Process
Filtering
Prediction
Importance sampling
On-line Monitoring
Markov processes
prediction
Empirical Bayes
Gibbs Sampling
Importance Sampling
Accidents
Noisy Data
Bayesian Methods
Spatial Data
nuclear accident
Resampling
Markov Chain Monte Carlo
Sampling

Keywords

  • Dynamic spatiotemporal process
  • Empirical Bayes estimation
  • Fukushima nuclear disaster
  • Geostatistics
  • Online inference
  • State space models

Cite this

Sequential Empirical Bayes Method for Filtering Dynamic Spatiotemporal Processes. / Evangelou, Evangelos; Maroulas, Vasileios.

In: Spatial Statistics, Vol. 21, No. A, 01.08.2017, p. 114-129.

Research output: Contribution to journalArticle

@article{288ff1ded8f4443e9ae6d050989049de,
title = "Sequential Empirical Bayes Method for Filtering Dynamic Spatiotemporal Processes",
abstract = "We consider online prediction of a latent dynamic spatiotemporal process and estimation of the associated model parameters based on noisy data. The problem is motivated by the analysis of spatial data arriving in real-time and the current parameter estimates and predictions are updated using the new data at a fixed computational cost. Estimation and prediction is performed within an empirical Bayes framework with the aid of Markov chain Monte Carlo samples. Samples for the latent spatial field are generated using a sampling importance resampling algorithm with a skewed-normal proposal and for the temporal parameters using Gibbs sampling with their full conditionals written in terms of sufficient quantities which are updated online. The spatial range parameter is estimated by a novel online implementation of an empirical Bayes method, called herein \emph{sequential empirical Bayes} method. A simulation study shows that our method gives similar results as an offline Bayesian method. We also find that the skewed-normal proposal improves over the traditional Gaussian proposal. The application of our method is demonstrated for online monitoring of radiation after the Fukushima nuclear accident.",
keywords = "Dynamic spatiotemporal process, Empirical Bayes estimation, Fukushima nuclear disaster, Geostatistics, Online inference, State space models",
author = "Evangelos Evangelou and Vasileios Maroulas",
year = "2017",
month = "8",
day = "1",
doi = "10.1016/j.spasta.2017.06.006",
language = "English",
volume = "21",
pages = "114--129",
journal = "Spatial Statistics",
issn = "2211-6753",
publisher = "Elsevier",
number = "A",

}

TY - JOUR

T1 - Sequential Empirical Bayes Method for Filtering Dynamic Spatiotemporal Processes

AU - Evangelou, Evangelos

AU - Maroulas, Vasileios

PY - 2017/8/1

Y1 - 2017/8/1

N2 - We consider online prediction of a latent dynamic spatiotemporal process and estimation of the associated model parameters based on noisy data. The problem is motivated by the analysis of spatial data arriving in real-time and the current parameter estimates and predictions are updated using the new data at a fixed computational cost. Estimation and prediction is performed within an empirical Bayes framework with the aid of Markov chain Monte Carlo samples. Samples for the latent spatial field are generated using a sampling importance resampling algorithm with a skewed-normal proposal and for the temporal parameters using Gibbs sampling with their full conditionals written in terms of sufficient quantities which are updated online. The spatial range parameter is estimated by a novel online implementation of an empirical Bayes method, called herein \emph{sequential empirical Bayes} method. A simulation study shows that our method gives similar results as an offline Bayesian method. We also find that the skewed-normal proposal improves over the traditional Gaussian proposal. The application of our method is demonstrated for online monitoring of radiation after the Fukushima nuclear accident.

AB - We consider online prediction of a latent dynamic spatiotemporal process and estimation of the associated model parameters based on noisy data. The problem is motivated by the analysis of spatial data arriving in real-time and the current parameter estimates and predictions are updated using the new data at a fixed computational cost. Estimation and prediction is performed within an empirical Bayes framework with the aid of Markov chain Monte Carlo samples. Samples for the latent spatial field are generated using a sampling importance resampling algorithm with a skewed-normal proposal and for the temporal parameters using Gibbs sampling with their full conditionals written in terms of sufficient quantities which are updated online. The spatial range parameter is estimated by a novel online implementation of an empirical Bayes method, called herein \emph{sequential empirical Bayes} method. A simulation study shows that our method gives similar results as an offline Bayesian method. We also find that the skewed-normal proposal improves over the traditional Gaussian proposal. The application of our method is demonstrated for online monitoring of radiation after the Fukushima nuclear accident.

KW - Dynamic spatiotemporal process

KW - Empirical Bayes estimation

KW - Fukushima nuclear disaster

KW - Geostatistics

KW - Online inference

KW - State space models

U2 - 10.1016/j.spasta.2017.06.006

DO - 10.1016/j.spasta.2017.06.006

M3 - Article

VL - 21

SP - 114

EP - 129

JO - Spatial Statistics

JF - Spatial Statistics

SN - 2211-6753

IS - A

ER -