R help - Feb 2002 - Best Subsets regression

Hi,

I have found a minor problem with leaps(). In 1.3.1 under Windows 2000 I 
seem to only be able to obtain values for one statistic at a time. That is 
choosing

method=c("Cp","adjr2","r2")

just gives Cp values.

To mimic the output of Minitab's

MTB > BReg 'Fertility'
'Agriculture'-'Infant.Mortality' ;
SUBC>   NVars 1 5;
SUBC>   Best 5;
SUBC>   Constant.

I need to go

data(swiss)
leapa <-leaps(as.matrix(swiss[,-1]),swiss[,1],method="Cp", nbest=5)
leapb <-leaps(as.matrix(swiss[,-1]),swiss[,1],method="adjr2",
nbest=5)
leapc <-leaps(as.matrix(swiss[,-1]),swiss[,1],method="r2", nbest=5)
best <- cbind(leapc$r2,leapb$adjr2,leapa$Cp,leapa$which)
colnames(best) <- c( "R.Sq",
"Adj.R.Sq","Cp","X1","X2","X3","X4","X5")
best

which seems rather wasteful, not that it matters on a small dataset like 
"swiss".

Actually I have the same grip with SPLUS, so I suppose this feature is 
retained for compatibility?

Dr Murray Jorgensen               on leave from:
Mathematics and Statistics       Department of Statistics
University of Victoria           University of Waikato
PO BOX 3045 STN CSC              Hamilton, New Zealand
Victoria, B.C.      Email: maj at waikato.ac.nz
Canada V8W 3P4      Phone: (250) 721-7460 wk  (250) 477 2581 home

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

On Wed, 13 Feb 2002, Murray Jorgensen wrote:
> Hi,
>
> I have found a minor problem with leaps(). In 1.3.1 under Windows 2000 I
> seem to only be able to obtain values for one statistic at a time. That is
> choosing
>
> method=c("Cp","adjr2","r2")
>
> just gives Cp values.
>
<snip>
>
> Actually I have the same grip with SPLUS, so I suppose this feature is
> retained for compatibility?
>
Yes.  It would be fairly easy to fix as all three statistics are computed.
The second through fourth lines from the end of leaps() select which
statistic to return.

However, leaps() is just there for backwards compatibility. The preferred
function is regsubsets(), and summary.regsubsets() does return all these
statistics (plus BIC). It also knows about model formulas and has a quite
nice plot() method.

	-thomas

Thomas Lumley			Asst. Professor, Biostatistics
tlumley at u.washington.edu	University of Washington, Seattle

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._