Håvard Rue & Haakon Bakka, from King Abdullah University of Science and Technology, will present the Department of Statistical Science seminar with a talk entitled, "An overview of the R-INLA project". 

Abstract: Most generalised linear mixed models (GLMMs), generalised additive models (GAMs), spline models, spatial models, and survival models, have one attribute in common: they contain a high-dimensional Gaussian distribution. In R-INLA (www.r-inla.org) we use fast numerical methods for sparse matrices to do approximate Bayesian inference for all of thesemodels quickly. Non-Gaussian likelihoods are included via the integrated nested Laplace approximation (INLA), and spatial models though the SPDE approach. The speed and accuracy obtained with the R-INLA package make even complicated and large models that was earlier considered as unpractical, now ready for routine use. A key example here is the class of log-Gaussian Cox processes. In this talk, we will review the basic ideas behind R-INLA, and discuss some applications towards ecology and joint survival models. We will also discuss some current work with the aim of constructing good models for non-separable space-time models based on time-dependent stochastic partial differential equations.