This course (12 hours) is provided by Prof. Marta Blangiardo (Imperial College London, U.K.) and Prof. Michela Cameletti (Università degli studi di Bergamo, Italy). They are the authors of the book “Spatial and Spatio-temporal Bayesian Models with R-INLA” published by Wiley.
Statisticians and applied researchers with strong interest in quantitative analysis.
The course will provide a focus on how to use the Integrated Nested Laplace Approximation approach (INLA) for the analysis of spatial and spatio-temporal data.
We will first go through the basic of INLA for Bayesian inference and will then see how to model hierarchical structures. We will then focus on area level data and present how to model spatially structured random effects through conditional autoregressive specifications; following that, we will extend the approach to include temporal dependency and touch briefly on spatio-temporal interactions. Moving on to point-referenced data we will introduce the stochastic partial differential equation (SPDE) approach, used for spatial modelling on a continuous field. We will then extend this to deal with spatio-temporal data. Finally we will describe how to use R-INLA for more advanced problems in the spatio-temporal realm, for instance how to deal with misaligned data or how to model two outcomes jointly.
The course will maintain a very practical angle, with a mixture of lectures and practical sessions. We will share a github repository with the material ahead of the course. For a book covering the topics of the course see https://onlinelibrary.wiley.com/doi/book/10.1002/9781118950203
It is recommended that people attending are familiar with R (https://www.r-project.org/) and with the basic of the Bayesian approach. It would be beneficial if attendees could bring their laptop with the latest version of R and INLA installed.
Wed 20th – morning (3 hours) Session 1.1: Lecture on introduction to INLA and R-INLA (1 hour) Session 1.2: Hierarchical models, prediction, prior specification (1 hour) Break Session 1.3: Practical session (1 hour)
Wed 20th – afternoon (3 hours) Session 1.4: Bayesian spatial model for small area data (1 hour) Session 1.5: Spatio-temporal modeling for small area data (1 hour) Break Session 1.6: Practical session (1 hour)
Thu 21st – morning (3 hours) Session 2.1: Models for geostatistical data and introduction to SPDE (1 hour) Session 2.2: inlabru (1 hour) Break Session 2.3: Practical session (1 hour)
Wed 21st – afternoon (3 hours) Session 2.4: Spatio-temporal model for geostatistical data (1 hour) Session 2.5 Advanced use of INLA (1 hour) Break Session 2.6: Practical session (1 hour)