[R] Best subset of models for glm.nb()

[Jari Oksanen]

Tim Clark <mudiver1200 <at> yahoo.com> writes:

> Yes, I have read about the problems with stepwise algorithms and am looking
for a valid alternative to
> narrowing down models when you have a lot of data and a large number of
variables your interested in.  
> 
Tim,

This is not an answer to your problem, but a comment. One of the popular sites
discussing problems with stepwise algorithms is

http://www.stata.com/support/faqs/stat/stepwise.html

which is based on Frank Harrell's comments. After listing ten major problems
with stepwise algorithms it adds:

"“All possible subsets” regression solves none of these problems."

Intriguing discussions may loom round the corner...

Cheers, Jari Oksanen