The model _is_ linear in parameters, after the log transformation of
the response, so
you don't need nlrq. If you really want something like:
y = exp(a + b x) + u
then you need to make a token effort to look at the documentation.
Here is another
example:
x <- exp(rnorm(50))
y <- exp(1 + .5*x) + rnorm(50)
nlrq(y ~ exp(a + b * x), start = list(a = 2, b = 1))
Nonlinear quantile regression
model: y ~ exp(a + b * x)
data: parent.frame
tau: 0.5
deviance: 15.39633
a b
1.0348673 0.4962638
Roger Koenker
rkoenker at illinois.edu
On Oct 16, 2011, at 3:59 AM, Julia Lira wrote:
>
> Dear all,
> I sent an email on Friday asking about nlrq {quantreg}, but I
> haven't received any answer.
> I need to estimate the quantile regression estimators of a model as:
> y = exp(b0+x'b1+u). The model is nonlinear in parameters, although I
> can linearise it by using log.When I write:
> fitnl <- nlrq(y ~ exp(x), tau=0.5)
> I have the following error: Error in match.call(func, call = cll) :
> invalid 'definition' argument
> Is there any way to estimate this model, or should I accept the
> following change:
> fitnl <- rq(log(y) ~ x, tau=0.5) ?
> Thanks in advance!
> Best,
> Julia
> [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.