[R] Problem with lme, ns and df (variable scoping problem?)

[Prof Brian Ripley]

Use substitute, something like

lml <- eval(substitute(lmList(y ~ ns(x,df=d.f) | c, a), list(d.f.=df)))

This sort of thing is in any case necessary to get a sensible call
component recorded in the fitted object.


On Thu, 22 Jan 2004, Hadley Wickham wrote:

> Hi,
> 
> I'm experimenting with random effect natural splines, and I've 
> encountered an odd problem.
> 
> library(nlme); library(splines)
> a <- data.frame(x = 1:10, y = 1:10 + runif(10, min=-3, max=3), c = 
> rep(c(1,2),5))
> df <- 10   
> lml <- lmList(y ~ ns(x,df=df) | c, a)
> 
> Error in df - 1 : non-numeric argument to binary operator
> 
> I presume this is because the formula isn't evaluated until it's deep in 
> the bowels of lmList, where a local df exists.  So I try again with a 
> variable name that probably doesn't exist elsewhere:
> 
> d.f <- 10
> lml <- lmList(y ~ ns(x,df=d.f) | c, a)
> 
> This works, but now when I calculate the mixed model, I get a different 
> error.
> 
> lme <- lme(lml, random=pdIdent(~c))
> Error in model.frame(formula, rownames, variables, varnames, extras, 
> extranames,  :
>         variable lengths differ
> 
> Neither of these problems occur if I specify df directly (eg. df=10)
> 
> Can anyone offer a work around?
> 
> Thanks.
> 
> Hadley
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
> 
> 

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
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