home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!wupost!usc!sdd.hp.com!uakari.primate.wisc.edu!zazen!schaefer.math.wisc.edu!horn
- From: horn@schaefer.math.wisc.edu (Jeffrey Horn)
- Newsgroups: sci.math.stat
- Subject: Best Subset with Non-Divisible Y-data
- Message-ID: <1992Jul21.175657.14090@schaefer.math.wisc.edu>
- Date: 21 Jul 92 17:56:57 GMT
- Organization: Univ. of Wisconsin Dept. of Mathematics
- Lines: 14
-
- I would like to do similar analysis to a "best subset search" for regression
- models (methods include regression by leaps and bounds, forward/backward
- selection, all subsets, etc.) for which is "non-divisible". By this I mean
- that one can rank the values of the independent and/or dependent variables
- from 1 to N, but a ranking of 2 is NOT TWICE BIG AS A ONE, etc. Some of
- the models I am looking at actually have mixtures of divisible and non-divisible
- independent variables. I am also looking at models in which the dependent
- variable is either non-divisible or divisible. I want to find something
- analogous to regression for these kinds of models. I am sure that any method
- that can be dug up will be less powerful than regression.
-
- Is there any method to handle such models?
-
- Jeff Horn
-