[R] nlme help please
Bill Simpson
wsi at gcal.ac.uk
Thu May 24 11:05:28 CEST 2001
I am trying to learn how to use nlme by working on a simple example. I
attach the data from a toy example I made up which is similar to my real
problem. (My grasp of fixed/random effects is still a bit tenuous)
It is a longitudinal study of the effect of two treatments: A and B. The
data were created by:
A: y<-12/(1+exp((2-time)/.5)),y<-8/(1+exp((2-time)/.5))
B: y<-22/(1+exp((2-time)/.5)), y<-18/(1+exp((2-time)/.5))
fixed effect for param a = 10,10, 20,20, random effects= +2, -2, +2, -2
fixed effect for param b=2, fixed effect for c = .5
normal noise with sd=1 was added to all y values.
I do:
df<-read.table(file="papers/alex/junk.dat",header=TRUE)
attach(df)
fullfit<-nlsList(y~SSlogis(Time,Asym,xmid,scal)|Subject,data=df)
and get the separate fits of the logistic to each subject's data. 4 curves
and 4 sets of param values (12 params total for the model).
How do I get the fit of the model which is the true one as stated above:
Asym for group A (fixed effect)
Asym for group B (fixed effect)
Asym diff from group A mean for subject 1 (random effect)
Asym diff from group A mean for subject 2 (random effect)
Asym diff from group B mean for subject 3 (random effect)
Asym diff from group B mean for subject 4 (random effect)
xmid (fixed effect)
scal (fixed effect)
= 8 params total
I would like 95% CIs for params and in particular a CI for the diff in
Asym for groups A and B (or a hypo test).
If I had an additional grouping factor (e.g. Sex), how would I do it? (I
would have enough subjects to make it worthwhile--I actually have 53).
Thanks very much for any help.
BTW I have not tried to use groupedData object because I think that's
mainly useful for the fancy trellis graphics, which we don't have in R. I
know there was a recent announcement about lattice, but I'm not at all
sure that would work with nlme at this point...
Bill Simpson
-------------- next part --------------
"Subject" "Group" "Time" "y"
1 a 1 1.803407
1 a 2 3.790964
1 a 3 6.885437
1 a 4 6.304922
1 a 5 6.423945
2 a 1 1.685487
2 a 2 4.775235
2 a 3 12.058397
2 a 4 12.500626
2 a 5 12.602189
3 b 1 1.532756
3 b 2 9.866473
3 b 3 16.928385
3 b 4 17.762473
3 b 5 18.493848
4 b 1 2.975398
4 b 2 13.158969
4 b 3 19.589508
4 b 4 22.389456
4 b 5 21.381210
More information about the R-help
mailing list