Hi!
I am a PhD student in EE from Germany with only limited background in
statistics.
My question is related to model selection:
Model selection techniques like AIC try to find the best fitting model
given a single sample. However, I want to compare the fit of a single
model (family), e.g. bivariate Gaussian, to different datasets. These
datasets are derived from a higher dimensional sample by different
(fixed) projections.
What methods (model selection criteria) are appropriate in that case?
Are likelihood based criteria allowed after a suitable normalization of
the samples?
I appreciate any help, thanks in advance!
Bernd Menser
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
