El professor Xavier Barber de la Universidad Miguel Hernández impartirá la conferència:"Modelling presence/absence under spatial misalignment using Bayesian latent Gaussian models"

Día: 10 de març de 2016,

Hora: 12:00

Lloc : Aula 0.5 de la Facultat de Matemàtiquess,

Resum: Modelling patterns of the presence/absence (species, diseases,...) using local
environmental factors has been a growing problem in the last few years. Geostatistical 
models have become popular lately because they allow to estimate and
predict the underlying disease risk and relate it with possible risk factors. Our
approach to these models is based on the fact that the presence/absence can be
expressed with a hierarchical Bayesian spatial model that incorporates the information
provided by the geographical and environmental characteristics of the region of interest. 
Nevertheless our main interest here is to tackle the misalignment
problem arising when information about possible covariates are partially
(or totally) different than those of the observed locations and those in which we
want to predict. As a result, we present two different models depending on the
fact that there is uncertainty on the covariates or not. In both cases, Bayesian
inference on the parameters and prediction of presence/absence in new locations
are made by considering the model as a latent Gaussian model, which allows
the use of the integrated nested Laplace approximation (INLA). In particular,
the spatial effect is implemented with the stochastic partial differential equation
(SPDE) approach. The methodology is evaluated on the presence of the fasciola
hepatica in Galicia, a North-West region of Spain.