[R] nls function
andrewr at uidaho.edu
Thu Apr 1 18:12:56 CEST 2004
I've just spent a few months trying to fit a model to a dataset, and it's not
easy. However, in my case, what appears to be recalcitrance on R's part
actually turns out to be well-founded warnings that the structure of the
model and the data are not permitting a clean, unambiguous fit.
I warmly recommend the Bates and Watts book on non-linear regression analysis,
and also Venables and Ripley MASS, with especial reference to the online
My strategy to learn more about my problem was: increase "tol" and decrease
minFactor until I get a fit, and then examine the intrinsic and parameter
curvatures using the rms.curv() function in the MASS package. It showed
quite clearly that there was a real problem with the parameterization of the
model. Of course, using that particular fit was out of the question.
I recommend against trying another tool: instead, learn more about this one
and why it's behaving in that way. Good luck.
On Thursday 01 April 2004 07:37, Bill Shipley wrote:
> Hello. I am trying to fit a non-rectangular hyperbola function to data
> of photosynthetic rate vs. light intensity. There are 4 parameters that
> have to be estimated. I find the nls function very difficult to use
> because it often fails to converge and then gives out cryptic error
> messages. I have tried playing with the control parameters but this
> does not always help.
> Is there another non-linear regression function in R that I might try
> (other than regression smoothers, which wont give the parameter
> estimates of the specified function)?
> Bill Shipley
> Subject Matter Editor, Ecology
> North American Editor, Annals of Botany
> Département de biologie, Université de Sherbrooke,
> Sherbrooke (Québec) J1K 2R1 CANADA
> Bill.Shipley at USherbrooke.ca
> [[alternative HTML version deleted]]
> R-help at stat.math.ethz.ch mailing list
> PLEASE do read the posting guide!
Andrew Robinson Ph: 208 885 7115
Department of Forest Resources Fa: 208 885 6226
University of Idaho E : andrewr at uidaho.edu
PO Box 441133 W : http://www.uidaho.edu/~andrewr
Moscow ID 83843 Or: http://www.biometrics.uidaho.edu
No statement above necessarily represents my employer's opinion.
More information about the R-help