This works in R-devel, for which NEWS says
o step(), add1.default() and drop1.default() now work somewhat
better if called from a function.
It's mainly a scoping problem, related to changes made way back in 1.2.x.
However, it is also an bad example, as z is really excluded in the fit.
stepAIC in the latest MASS (VR_7.0-11) will work, but warns about this.
I hope that this is just an over-simplification of the real problem.
On Sun, 2 Mar 2003, Darryl Greig wrote:
> Could someone point me in the right direction for the following issue:
>
> A function is defined as follows:
>
> tfun <- function(dat)
> {
> fmla <- as.formula("y~x+z")
> dat2 <- dat
> mdl <- lm(fmla,dat2)
> mdl <- step(mdl)
> }
>
> Then the following code
>
> dat <- data.frame(x=1:10,z=1:10,y=(1:10)^2+10*(1:10))
> tfun(dat)
>
> generates the output
>
> Start: AIC= 43.67
> y ~ x + z
>
> Error in model.frame.default(formula = y ~ z, data = dat2,
> drop.unused.levels = TRUE) :
> Object "dat2" not found
>
> Any help or pointers gratefully accepted.
>
> Thanks,
> Darryl Greig ([EMAIL PROTECTED])
>
> ______________________________________________
> [EMAIL PROTECTED] mailing list
> http://www.stat.math.ethz.ch/mailman/listinfo/r-help
>
--
Brian D. Ripley, [EMAIL PROTECTED]
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UK Fax: +44 1865 272595
______________________________________________
[EMAIL PROTECTED] mailing list
http://www.stat.math.ethz.ch/mailman/listinfo/r-help
