Short courses

Abstract: This short course will provide an introduction to the theory and methods of model-based geostatistics and their implementation in the R contributed package geoR, together with a selection of environmental and health sciences applications.

Geostatistics refers to the sub-branch of spatial statistics in which the data consist of a finite sample of measured values derived from an underlying spatially continuous phenomenon. Many of the ideas and methods of classical geostatistics were developed by the Fontainebleau school, largely independently of the mainstream of spatial statistics. One consequence of this was a distinctive terminology and style which tended to conceal the strong connections with parallel developments in spatial statistics. Model-based geostatistics is an approach to geostatistical problems based on the application of formal statistical methods under an explicitly assumed stochastic model.

At the end of the course, students should understand the following: the relationship of classical geostatistics to mainstream spatial statistics, the theory and application of likelihood-based and Bayesian inference for geostatistical models and data, the relevance of geostatistical models and methods to substantive problems in the environmental and health sciences, and the implementation of model-based geostatistics using the geoR contributed package within the R computing environment.

Much of the material in the course, including the software demonstrations, is taken from the newly published text:

DIGGLE, P.J. and RIBEIRO, P.J. (2007). Model-based Geostatistics. New York : Springer.

Abstract: This strictly introductory course in Bayesian modelling and analysis will comprise a mix of theory, methods and computation, presented as a merger of tutorials and hands-on practicals. The course will cover an overview of the Bayesian approach to modelling, Bayesian computation using Markov chain Monte Carlo and WinBugs, and issues in Bayesian modelling. Two case studies in hierarchical and latent variable models, using meta-analysis and mixtures as concrete examples, will consolidate the workshop ideas. The third case study will focus on noninformative and informative priors, and will include a brief discussion of Bayesian networks.

The short course registration fees have been set as follows:

IBS Member	$225
Non-member	$275
Student IBS member	$150
Student Non-member	$200

Your registration fee includes morning and afternoon teas, lunch and any relevant handouts.