[R] specifying starting values in nlsList/conversion problems

Schmidt.Matthias (FORST) Matthias.Schmidt at forst.bwl.de
Thu May 27 15:10:00 CEST 2004


Hi 
I am using nlsList from the nlme package to estimate group specific
coefficient sets to get a first impression of the structure of coefficients
in relation to potential predictor variables. In some cases I got
coefficients in other there are conversion problems due to bad start values
I guess, resulting in typical error messages as known from nls:

Error in nls(formula = formula, data = data, start = start, control =
control) : 
        singular gradient
Error in nls(formula = formula, data = data, start = start, control =
control) : 
        step factor 0.000488281 reduced below `minFactor' of 0.000976563
Error in numericDeriv(form[[3]], names(ind), env) : 
        Missing value or an Infinity produced when evaluating the model

is there a possibility to specifiy group specific starting or an algorithm
which tests a matrix of starting with a certain stepwidth to overcome the
conversion problems?

thanx for help

***********************************************************
Matthias Schmidt
Forstliche Versuchs- und Forschungsanstalt Baden-Württemberg (FVA) 
Abteilung Biometrie und Informatik
Wonnhaldestr. 4
79100 Freiburg i. Br
Tel.: + 49 (0)761 / 4018 -187
Fax: + 49 (0)761 / 4018 - 333




More information about the R-help mailing list