|
Title
|
Room
|
|
Statistical Issues in Spatial Modeling of Small Datasets
|
H-Gwinnett
|
|
Date / Time
|
Sponsor
|
Type
|
|
08/08/2001
8:30 AM
-
10:20 AM
|
ENAR
|
Invited
|
|
Organizer:
|
Mary Christman, University of Maryland
|
|
Chair:
|
Mary Christman, University of Maryland
|
|
Discussant:
|
|
|
Floor Discussion
|
10:15 AM
|
Description
Spatial data analysis is a set of relatively recent modeling techniques which provide for analysis and modeling in the presence of correlated data structures. The form of the correlation typically describes a purely two (occasionally three)-dimensional spatial autocorrelation or adds a temporal component. Methods have been based on assumptions about stationarity, isotropy, and on normality of the data. As a consequence, typical mean fields and autocorrelation structures are highly constrained. Problems with modeling spatial data include such issues as incorporating covariates into the mean field, dealing with obvious nonstationarity and anisotropic issues, inferences for data which are not normally distributed, combining data measured on different scales,and other similar problems. This session presents a few of the problems encountered when analyzing small to medium sized dataset and some recommended solutions.
|
|
