Extending JAGS for spatial data

• Ano de defesa: 2018
• Autor(a) principal: Magno Tairone de Freitas Severino
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Jags; Amostrador de gibbs; Inferência bayesiana; Estatística espacial; Estatistica; Teoria bayesiana de decisão estatistica; Análise espacial (Estatística)
• Link de acesso: http://hdl.handle.net/1843/BUOS-AZGNHV

Bayesian hierarchical modeling for spatial data is challenging for professionals from other areas than statistics. From a technical perspective, setting the model and the prior distributions are the simplest part of the process. What makes it dicult is the com putation of the posterior full conditionals and the implementation of the Gibbs Sampleralgorithm. The BUGS (Bayesian inference Using Gibbs Sampling) family of statistical softwares reduces the eort of modeling, since the user must indicate only the prior distributions and the likelihood function. However, in general these softwares do not im plement several spatial models, although users of WinBUGS and OpenBUGS can enjoy fromthe spatial add-on called GeoBUGS. JAGS (Just Another Gibbs Sampler), the open-source C++ developed version of the BUGS family, does not contain any function or distribution for spatial modeling. This project aims to ll this gap through the implementation of an extension to the JAGS software, allowing users from dierent elds to perform a spatialdata modeling and analysis