[R-sig-ME] Data weighting in gnls

[Nika Galic]

Dear all,

I have a data set in which different species contribute different numbers
of data points and I am fitting non-liner models to it. Does anyone know
how I could weight each data point so that species with few points are
equally weighted as those with many data points?

I would assume that something where each point is weighted by 1/n (n-number
of data points for a single species) for the model fitting routine could be
the answer, but I am not totally sure if this is the way and how to
implement this. I am using the *gnls *function in the nlme package, and am
applying a phylogenetic correction (using the correlation argument) for
model fitting.

Any suggestions are very much appreciated, apologies if this group is not
the most appropriate for this posting.

Thanks,
Nika

	[[alternative HTML version deleted]]