Short Course on Bayesian Modeling and Inference for HighDimensional SpatialTemporal Data
Abhi Datta (Johns Hopkins University), Sudipto Banerjee (UCLA), Andrew O. Finley (Michigan State University)
In this course, we will present scalable Bayesian models and related Markov chain Monte Carlo based methods that can provide fast analysis of big spatial and spatiotemporal data using modest computing resources. We will begin with an introduction on pointreferenced spatial data accompanied by an overview of statistical methods used to analyze them. We will briefly cover exploratory data analysis techniques like variogram fitting, basics of geo statistical approaches like kriging and Gaussian Processes, and fundamentals of Bayesian hierarchical geospatial models. We will then highlight some of the computational issues experienced by Gaussian Process models when used to model large spatial data. In this context, we will present new scalable Bayesian models that can deliver fully model based inference for massive spatial data and demostrate how to analyze large datasets using these models in R.
Course Schedule

Introduction to Pointreferenced Spatial Data (8:30 AM – 10:15 AM)

Introduction to Geostatistics
(3x2 slides)

Bayesian Conjugate Linear Regression
(3x2 slides)

Spatial mapping and spBayes/R

BREAK: 10:15AM10:30AM

Highdimensional Bayesian Geostatistics (10:30AM12:30PM)

Lowrank models and the Predictive Process
(3x2 slides)

Sparsity composite likelihoods and the Nearest Neighbor Gaussian Process (NNGP)
(3x2 slides)

Spatiotemporal NNGP
(3x2 slides)

LUNCH: 12:30PM2:00PM

Spatial Analysis in R using Predictive Processes and NNGPs (2:00PM3:15PM)

Advanced computing environments
(3x2 slides)

spBayes demo for Predictive Process

spNNGP Intro and Demo

BREAK: 3:15PM3:30PM

Practical Computational Considerations (3:30PM5:00PM)

Hierarchical Modeling for Large Univariate Areal Data
(3x2 slides)