|
Title
|
Room
|
|
! Robust Statistics for Correlated Data
|
H-State/Club
|
|
Date / Time
|
Sponsor
|
Type
|
|
08/07/2001
8:30 AM
-
10:20 AM
|
Section on Statistics & the Environment*, ENAR
|
Invited
|
|
Organizer:
|
Marc Genton, North Carolina State University
|
|
Chair:
|
Montserrat Fuentes, North Carolina State University
|
|
Discussant:
|
|
|
Floor Discussion
|
10:15 AM
|
Description
Over the last decades, robust statistics has been mainly involved with uncorrelated data. Roughly speaking, robust statistics is concerned with the fact that many assumptions commonly made in statistical methods are at most approximations of the reality. This might be particularly true for correlated data sets, for example arising from Environmental Sciences, where it is difficult to detect and deal with outlying values. Some attempts have been made to address these issues for time series, but very few in the context of space or space-time data. The challenges for research statisticians on this topic will be to define the notion of "outlying values" for correlated data; construct new tools to characterize the robustness of statistical methods in the presence of correlation; deal with large to massive correlated data sets. The goal of this session is to open up new perspectives on the topic of robustness for correlated data and foster interactions between statisticians interested in Environmental Sciences and those involved with robust statistics.
|
|
