Andras Zempleni: Statistical modeling in moderately high dimensions with emphasis on extremes

Andras Zempleni (Eötvös Loránd University Budapest) skal snakke om

Statistical modeling in moderately high dimensions with emphasis on extremes

Sammendrag

With the quickly increasing computing power and the developing modeling frameworks, multivariate statistical modeling also became feasible. However, most of the approaches are suitable only for low dimensional cases. There is one exception, the vine copulas were purposely introduced for high dimensional problems.

In this talk I investigate the possible approaches, from an extreme-value perspective: copulas and different extreme-value models will be analysed and applied to 5-dimensional German wind data.

This is joint work with Pál Rakonczai.

Published Jan. 12, 2012 11:16 AM - Last modified Apr. 16, 2012 1:47 PM