Marieke
In addition to Nick's observations, you still need the commands
matrix B=e(b)
local finvars : colfullnames(B)
after the -sw logit- command in -myboot- and -local vinf ...- should be inside the foreach loop, making it:
foreach v of varlist age bmi pp {
local vinf : list v in finvars
return scalar `v'=`vinf'
}
Note again that in version 8 it is important that nothing follows the -{- on the -foreach- line and that the -}- is on a line of its own.
You could also drop -r(rep78)- from the -bootstrap- command as this refers to a variable in auto.dta, but this probably won't stop it working.
Finally, -saving()- should be given as an option of -bootstrap-, not -save- as a separate command, and the -replace- option should also not be in the brackets of -saving()-, ie
bootstrap ... saving(bootresults) replace
Hope this helps
David
[email protected]
-----Original Message-----
From: Nick Cox [mailto:[email protected]]
Sent: 21 July 2004 15:33
To: [email protected]
Subject: RE: st: obtaining AUC at each iteration when bootstrapping a
logistic reg ression model
Your -foreach- loop is at a minimum 3 lines,
with "foreach" as one word and { } aligned
as here.
foreach v of varlist age bmi pp {
...
}
I don't understand the line in the middle, but
this is a start.
Nick
[email protected]
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of
> [email protected]
> Sent: 21 July 2004 15:25
> To: [email protected]
> Subject: RE: st: obtaining AUC at each iteration when bootstrapping a
> logistic reg ression model
>
>
> David and Nick,
>
> Thanks for your help!
>
> While running the do file I encountered the following error message:
>
> command -> myboot
> error when command executed on original dataset under
> version 8 control
> r(198);
>
> By now, it is probably quite obvious to you that I am just
> beginning to
> learn to work with Stata, so it might be something very
> silly. Still, I hope
> you will help me.
>
> This is the do-file (as applied to a (simplified version of)
> my data set (my
> model has of course more independent variables)):
>
> capture program drop myboot
> program myboot, rclass
> sw logistic ras age bmi pp, forward pe (.1)
> local vinf : list v in finvars
> lroc, nograph
> return scalar area=r(area)
> for each v of varlist age bmi pp {return scalar `v'= (
> `vinf' " , "
> `v' " )>0}
> end
>
> clear
> use "C:\My documents\prediction model.dta"
> sw logistic ras age bmi pp, forward pe(.1)
> lroc, nograph
> bootstrap "myboot" r(area) r(rep78) r(age) r(bmi)
> r(pp), reps(100)
> save (bootresults, replace)
>
> clear
> use bootresults
> list
>
> This is the output:
>
> . capture program drop myboot
>
> . program myboot, rclass
> 1. sw logistic ras age bmi pp, forward pe (.1)
> 2. local vinf : list v in finvars
> 3. lroc, nograph
> 4. return scalar area=r(area)
> 5. for each v of varlist age bmi pp {return scalar `v'= (
> `vinf' " , " `v'
> " )>0}
> 6. end
>
> .
> . clear
>
> . use "C:\My documents\prediction model.dta"
>
> . sw logistic ras age bmi pp, forward pe(.1)
> begin with empty model
> p = 0.0001 < 0.1000 adding age
> p = 0.0056 < 0.1000 adding bmi
>
> Logistic regression Number of obs =
> 328
> LR chi2(2) =
> 26.55
> Prob > chi2 =
> 0.0000
> Log likelihood = -154.12872 Pseudo R2 =
> 0.0793
>
> --------------------------------------------------------------
> --------------
> --
> ras | Odds Ratio Std. Err. z P>|z| [95% Conf.
> Interval]
> -------------+------------------------------------------------
> --------------
> --
> age | 1.052812 .0135307 4.00 0.000 1.026624
> 1.079669
> bmi | .909435 .0311763 -2.77 0.006 .8503382
> .972639
> --------------------------------------------------------------
> --------------
> --
>
> . lroc, nograph
>
> Logistic model for ras
>
> number of observations = 328
> area under ROC curve = 0.6926
>
> . bootstrap "myboot" r(area) r(rep78) r(age) r(bmi) r(pp), reps(100)
>
> command -> myboot
> error when command executed on original dataset under version
> 8 control
> r(198);
>
> end of do-file
> r(198);
>
> Many thanks in advance,
>
> Marieke
> [email protected]
>
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
