Bayesian spatial models in INLA

R-INLA can fit a broad class of generalized linear mixed models (GLMMs), including spatial and spatio-temporal models: those that can be expressed as latent Gaussian Markov random fields (GMRF).

In bayesian modelling, a posterior distribution is estimated for every model parameter.

INLA algorithm (Integrated Nested Laplace Approximation): fast approximation method to obtain the posterior distributions; very fast compared to MCMC approximation.

R-INLA provides several criteria to compare and evaluate models (DIC, WAIC, CPO).