I'm trying to fit my first nlmer model. I follow Doug Bates's code at http://lme4.r-forge.r-project.org/slides/2009-07-01-Lausanne/8NLMMD.pdf and use lme4_0.999375-39.
A call to nls works fine, but when I turn it into a call to nlmer and add the random component, errors are returned. Providing start as a named vector returns one kind of error; providing start as a list returns a different error.
My data are simulated but they contain noise (per the warning on the nlmer help page).
You can see a plot of the data at http://jacobwegelin.net/r-help/2011_03_27_1311_simulated.pdf.
The error messages are as follows. Code to reproduce the entire problem is below under Details.
> roop <-nlmer(
+ Y ~ xintercept + amplitude * sin( frequency * X - xshift )
+ ~ (xshift | ID)
+ , data=mdoot
+ , start=list(xintercept=1, amplitude=1.5, frequency=1, xshift=pi/2)
+ )
Error: length(start$fixef) > 0 is not TRUE
>
> roop <-nlmer(
+ Y ~ xintercept + amplitude * sin( frequency * X - xshift )
+ ~ (xshift | ID)
+ , data=mdoot
+ , start=c(xintercept=1, amplitude=1.5, frequency=1, xshift=pi/2)
+ )
Error in nlmer(Y ~ xintercept + amplitude * sin(frequency * X - xshift) ~ :
gradient attribute of evaluated model must be a numeric matrix
>
Details:
set.seed(1)
N<-100; nIDs<-20
X<-seq(1,19, length=N)
Y<-sin(X) + rnorm(N)/10
IDs<-1:nIDs
trueRandomEffects<-data.frame(
xintercept=rnorm(nIDs)/3
, amplitude=1+ rnorm(nIDs)/9
, frequency =1+rnorm(nIDs)/10
, xshift =runif(nIDs) * pi
)
rownames( trueRandomEffects)<- IDs
mdoot<- DAT<-data.frame(ID=IDs[1], X=X, Y=Y)
for(thisperson in IDs[-1]) {
nextOne<-DAT
nextOne$Y<- (
trueRandomEffects[thisperson,"xintercept"]
+ trueRandomEffects[thisperson,"amplitude"]
* sin(
trueRandomEffects[thisperson,"frequency"] * X
- trueRandomEffects[thisperson,"xshift"]
)
+ rnorm(N)/10
)
nextOne$ID<- thisperson
mdoot<-rbind( mdoot, nextOne)
}
attr(mdoot, "ranef")<-trueRandomEffects
require(ggplot2)
print(
ggplot(data=mdoot, mapping=aes(x=X, y=Y, group=ID, color=ID)) + geom_path()
)
whump<-nls(
Y ~ xintercept + amplitude * sin( frequency * X - xshift
)
, data=mdoot
, start=list(xintercept=1, amplitude=1.5, frequency=1, xshift=pi/2)
)
print(whump)
print(summary(whump))
mdoot$nlsfit<-predict(whump)
print(head(mdoot))
print(
ggplot(data=mdoot, mapping=aes(x=X, y=Y, group=ID, color=ID)) + geom_path() + geom_point(mapping=aes(x=X, y=nlsfit))
)
# Now where it stops working:
require(lme4)
roop <-nlmer(
Y ~ xintercept + amplitude * sin( frequency * X - xshift )
~ (xshift | ID)
, data=mdoot
, start=list(xintercept=1, amplitude=1.5, frequency=1, xshift=pi/2)
)
roop <-nlmer(
Y ~ xintercept + amplitude * sin( frequency * X - xshift )
~ (xshift | ID)
, data=mdoot
, start=c(xintercept=1, amplitude=1.5, frequency=1, xshift=pi/2)
)
> sessionInfo()
R version 2.12.2 (2011-02-25)
Platform: i386-apple-darwin9.8.0/i386 (32-bit)
locale:
[1] en_US.UTF-8/en_US.UTF-8/C/C/en_US.UTF-8/en_US.UTF-8
attached base packages:
[1] grid stats graphics grDevices utils datasets methods
base
other attached packages:
[1] lme4_0.999375-39 Matrix_0.999375-46 lattice_0.19-17
ggplot2_0.8.8 proto_0.3-8
[6] reshape_0.8.3 plyr_1.2.1
loaded via a namespace (and not attached):
[1] digest_0.4.2 nlme_3.1-98 stats4_2.12.2
>
Thanks for any insights
Jacob A. Wegelin
Assistant Professor
Department of Biostatistics
Virginia Commonwealth University
730 East Broad Street Room 3006
P. O. Box 980032
Richmond VA 23298-0032
U.S.A.
E-mail: jacobwegelin at fastmail.fm
URL: http://www.people.vcu.edu/~jwegelin