In this 1-day course, we discuss approximate Bayesian inference for a class of models named "latent Gaussian models" (LGM). LGM's are commonly used class of models in statistical applications, including most of (generalized) linear models, (generalized) additive models, smoothing spline models, state space models, semiparametric regression, spatial and spatiotemporal models, log-Gaussian Cox processes and geostatistical and geoadditive models. Our approach to (approximate) Bayesian inference is to use integrated nested Laplace approximations (INLA). Their main benefit is computational: where MCMC algorithms need hours or days to run, these approximations provide more precise estimates in seconds or minutes. In this short course the background for understanding LGM and INLA will be introduced; why it works and why its fast. We will end this lecture illustrating INLA on some examples in R. Please visit www.r-inla.org to download the package and for further documentation.
Course fee: Free (need to register in w3.math.uminho.pt/STATMOD/registration.html, maximum attendees 30 according to registration order)
|