[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Fwd: Re: [R] regsubsets selection criterion]

If you did not join the r-help list, you are missing interesting stuff. Here's one that really fascinates me. All automated model selection regimes lead to the same answers with the same cases. Humphf. Who'd have thunk it?
Now that the lab computers have all buildable packages installed, I think you could test out this leaps thing and let me know how you like it. I never had much faith in these things, but some of the real scientists seem to. 

--
Paul E. Johnson                       email: pauljohn_AT_ku.edu
Dept. of Political Science            http://lark.cc.ku.edu/~pauljohn
1541 Lilac Lane, Rm 504
University of Kansas                  Office: (785) 864-9086
Lawrence, Kansas 66044-3177           FAX: (785) 864-5700
--- Begin Message --- 
On Tue, 27 Sep 2005, Samuel Bertrand wrote:

> Hello,
>
> I am using the 'regsubsets' function
> (from leaps package)
> to get the best linear models
> to explain 1 variable
> from 1 to 5 explanatory variables
> (exhaustive search).
>
> Is there anyone who can tell me
> on which criterion is based
> the 'regsubsets' function ?
>

As you get a best model for each size, all the commonly used criteria are 
equivalent. RSS, AIC, BIC, Mallows Cp, adjusted R^2, etc, all give the 
same ordering for models with the same number of predictors.


 	-thomas

______________________________________________
R-help_AT_stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html

--- End Message ---