Re: Multinomial estimation in (non-random) subsets

From: George Kahrimanis (anakreon_at_hol.gr)
Date: 11/19/04 

Date: Fri, 19 Nov 2004 19:45:27 +0200
To: Markus Mottl <markus@fichte.ai.univie.ac.at>

Marcus Mottl asked

>Does anybody know of a statistically sound and efficient way of
>incorporating the knowledge about the probability distribution of
>the total set to estimate a probability distribution of a (possibly
>biased) subset?

If you do not restrict what do you mean by "biased", we are totally
clueless on the pdf of the subset. E.g., you can model the actions
of the selector.

~ George Kahrimanis