[R] lattice: showing panels for factor levels with no values

Deepayan Sarkar deepayan at stat.wisc.edu
Fri Feb 13 18:22:57 CET 2004


On Friday 13 February 2004 07:18, Dieter Menne wrote:

[...]

> Dieter Menne:
>
> See also
>
> http://maths.newcastle.edu.au/~rking/R/help/01c/2820.html
>
> Deepayan, at that time I received an e-mail (not in the archives) from you
> saying that this feature was already in the code, but not yet active
> (quoted from memory).

I remember seeing this plot, but I can't remember (or imagine) what solution I 
was thinking of. In your example, the number of levels is already 17 in the 
fit returned by lme():

> fm4Oats <- lme( yield ~ nitro, data = Oats1, random = ~ 1 | Block/Variety ) 
> fm4Oats
Linear mixed-effects model fit by REML
  Data: Oats1 
[...]
Number of Observations: 68
Number of Groups: 
             Block Variety %in% Block 
                 6                 17 

and the subsequent augPred and plot.augPred calls retain that. In other words, 
the 'dropping' of levels has already happened when xyplot is called, and the 
fact that xyplot would have dropped unused levels doesn't come into play. I 
don't think there's an easy way to get what you want here.

Deepayan




More information about the R-help mailing list