[R-sig-ME] nlmer model definition and starting values

Sebastian P. Luque spluque at gmail.com
Wed Apr 29 15:07:23 CEST 2009

On Wed, 29 Apr 2009 09:52:31 +0200,
"ONKELINX, Thierry" <Thierry.ONKELINX at inbo.be> wrote:

> Dear Sebastian, There is no need to define SSGompertz as SSgompertz is
> already included in the stats package.

> The helpfile for nlme() gives you a good example. Switching SSamymp
> into SSgompertz is rather obvious.

> fm1 <- nlme(height ~ SSasymp(age, Asym, R0, lrc), data = Loblolly,
> fixed = Asym + R0 + lrc ~ 1, random = Asym ~ 1, start = c(Asym = 103,
> R0 = -8.5, lrc = -3.3)) summary(fm1) fm2 <- update(fm1, random =
> pdDiag(Asym + lrc ~ 1)) summary(fm2)

Thanks Thierry.  I was aware of SSgompertz(), but it is not the
parameterization that I need, and I don't know how I would have to
back-transform its results so that they correspond to the model I
showed.  I was also not sure how to define the call to nlmer() with the
fixed and random effects I described, and particularly how to set up the
start values.  The latter needs a named list or a vector, but I have
trouble anticipating what those names should be or the order of elements
in such a vector.



More information about the R-sig-mixed-models mailing list