 
A oneday handson workshop is always held in advance of
the Graybill Conference.
The workshop will be held on the Colorado State University campus in room 205 of the Weber Building. Click here for a map. The workshop will begin at 9am, lunch on your own at noon, then it resumes at 1:30pm and concludes at 4pm.
An introduction to the analysis of extreme values using R and extRemes
The instructors for the workshop will be:
Eric Gilleland, NCAR, Boulder CO
Mathieu Ribatet, EPFL, Lausanne, Switzerland
Description for the first part of the workshop:
Much of the statistical analysis tools taught in introductory statistics courses
concern with the center of mass of distributions, often relying on the Central
Limit Theorem (CLT) as justification for assuming that a sample of data follow
the Normal distribution. When interest is in extremes, however, much of
the data can be less useful, and the assumption of normality may not be
appropriate. A similar theorem to the CLT, the Extremal Types Theorem,
provides justification for the generalized extreme value (GEV) family of
distributions under certain assumptions.
This workshop will give some background on extreme value analysis (EVA), and an
introduction to fitting data to the GEV as well as threshold excess models (the
generalized Pareto (GP) and point process). The R programming language
will be used, but no knowledge of the language is required as the graphical user
interface (GUI) R package, extRemes, will primarily be used. Some
instruction on using R will also be given to assist in going beyond the
capabilities of the GUI's.
In addition to the introduction to statistical analysis of extremes, the
workshop will present a tutorial on spatial extremes using the spatial
extremes package in R. A description of the package follows:
The SpatialExtremes package aims to provide tools for modeling spatial extremes.
Many environmental processes are spatial by nature and the knowledge of the
univariate/multivariate distribution of extremes is not enough. This package
aims to fill this gap by using several approaches. Currently, the modeling is
performed through the maxstable framework. Maxstable processes are the
extension of the extreme value theory to random fields. Later, different
approaches will be implemented such as the latent variable and the copula based
methodologies.
