Open Access
Translator Disclaimer
February 2017 Inference on dynamic models for non-Gaussian random fields using INLA
R. X. Cortes, T. G. Martins, M. O. Prates, B. A. Silva
Braz. J. Probab. Stat. 31(1): 1-23 (February 2017). DOI: 10.1214/15-BJPS300

Abstract

Robust time series analysis is an important subject in statistical modeling. Models based on Gaussian distribution are sensitive to outliers, which may imply in a significant degradation in estimation performance as well as in prediction accuracy. State-space models, also referred as Dynamic Models, is a very useful way to describe the evolution of a time series variable through a structured latent evolution system. Integrated Nested Laplace Approximation (INLA) is a recent approach proposed to perform fast approximate Bayesian inference in Latent Gaussian Models which naturally comprises Dynamic Models. We present how to perform fast and accurate non-Gaussian dynamic modeling with INLA and show how these models can provide a more robust time series analysis when compared with standard dynamic models based on Gaussian distributions. We formalize the framework used to fit complex non-Gaussian space-state models using the R package INLA and illustrate our approach with a simulation study and a Brazilian homicide rate dataset.

Citation

Download Citation

R. X. Cortes. T. G. Martins. M. O. Prates. B. A. Silva. "Inference on dynamic models for non-Gaussian random fields using INLA." Braz. J. Probab. Stat. 31 (1) 1 - 23, February 2017. https://doi.org/10.1214/15-BJPS300

Information

Received: 1 February 2015; Accepted: 1 September 2015; Published: February 2017
First available in Project Euclid: 25 January 2017

zbMATH: 1362.62054
MathSciNet: MR3601658
Digital Object Identifier: 10.1214/15-BJPS300

Keywords: Approximate Bayesian inference , dynamic models , homicide rates , INLA , MCMC

Rights: Copyright © 2017 Brazilian Statistical Association

JOURNAL ARTICLE
23 PAGES


SHARE
Vol.31 • No. 1 • February 2017
Back to Top