UTAH CHAPTER
American Statistical Association
 
2007 Annual Southern Regional Meeting
 
Thursday, November 1, 2007
 
4:00pm
1170 TMCB
Brigham Young University
 
 
“Prediction for Max-Stable Processes via
an Approximated Conditional Density”
 
Dan S. Cooley
Department of Statistics
Colorado State University
 
Abstract:
 
The dependence structure of a max-stable random vector is characterized by its spectral measure. Given only the spectral measure, we present a method for approximating the conditional density of an unobserved component of a max-stable random vector given the other components of the vector. The approximated conditional density can be used for prediction. We also present a new parametric model for the spectral measure of a multivariate max-stable distribution. This model is used to perform prediction for both a time series and spatial process.