ONKELINX, Thierry <Thierry.ONKELINX <at> inbo.be> writes:
> This solutions works with R 2.7.0 under windows
>
> library(MASS)
> library(nlme)
> PredRes <- function(cal, val){
> cal <<- cal
> lmemod <- lme(distance ~ age * Sex, random = ~1|Subject, data = cal,
> method="ML")
> themod <- stepAIC(lmemod, dir="both", trace = FALSE)
> prs <- predict(themod, newdata = val)
> obs <- val$distance
> print(mean(obs - prs))
> }
>
> PredRes(cal = subset(Orthodont, age!=14), val = subset(Orthodont,
> age==14))
> obs<-subset(Orthodont,age==14)$distance
> > print(mean(obs-prs))
That's a workaround, but it effectively makes cal global, and so should be
considered not so nice programming practice. The problem is an environment
problem in stepAIC, and I am sure Brian Ripley will tell us what's wrong...
10: inherits(x, "data.frame")
9: is.data.frame(data)
8: lme.formula(fixed = distance ~ age + Sex, data = D1, random = ~1 |
Subject, subset = age != 14, method = "ML")
7: eval(expr, envir, enclos)
6: eval(expr, p)
5: eval.parent(nfit)
4: dropterm.default(fit, scope$drop, scale = scale, trace = max(0,
trace - 1), k = k, ...)
3: dropterm(fit, scope$drop, scale = scale, trace = max(0, trace -
1), k = k, ...)
2: stepAIC(lmemod, dir = "both")
1: PredRes(Orthodont)
R version 2.8.0 Under development (unstable) (2008-05-03 r45596)
i386-pc-mingw32
locale:
LC_COLLATE=German_Germany.1252;LC_CTYPE=German_Germany.1252;
LC_MONETARY=German_Germany.1252;LC_NUMERIC=C;LC_TIME=German_Germany.1252
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] nlme_3.1-88 MASS_7.2-42
loaded via a namespace (and not attached):
[1] grid_2.8.0 lattice_0.17-7 tools_2.8.0
>
Dieter
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
