[R] Re: nlrq problem

Roger Koenker roger at ysidro.econ.uiuc.edu
Thu Nov 20 15:26:29 CET 2003


You can minimize an model expression by just putting the ~ on the left
and everything else on the righthand side, but I don't think that this
is really what you want.  In the NLS expression this would ignore the
jacobian of the transformation from errors to response, and in nlrq
there is the same problem, however you can adjust for the jacobian
by rescaling by the geometric mean of the response.  I hope that this


url:	www.econ.uiuc.edu/~roger/my.html	Roger Koenker
email	rkoenker at uiuc.edu			Department of Economics
vox: 	217-333-4558				University of Illinois
fax:   	217-244-6678				Champaign, IL 61820

On Thu, 20 Nov 2003, Johannes Ludsteck wrote:

> Dear Mr. Koenker,
> sorry for bothering you. I have a probably trivial problem with
> your nlrq procedure.
> I would like to estimate the simple box-cox tranformed model
>  nlrq(I(y^t-1)/t) ~ I(a+b*x), data=df, start = c(t=1,a=0,b=0))
> R returns with the error message
>   Error in unique(c("AsIs", oldClass(x))) : Object "t" not found
> Is nlrq not able to handle transformations of the left hand side
> variable or have I maked a mistake?
> Thanks for every help and best regards,
> 	Johannes Ludsteck
> <><><><><><><><><><><><>
> Johannes Ludsteck
> Economics Department
> University of Regensburg
> Universitaetsstrasse 31
> 93053 Regensburg
> Phone +49/0941/943-2741

More information about the R-help mailing list