R help - Jul 2009 - Making rq and bootcov play nice

I have a quick question, and I apologize in advance if, in asking, I
expose my woeful ignorance of R and its packages. I am trying to use
the bootcov function to estimate the standard errors for some
regression quantiles using a cluster bootstrap. However, it seems that
bootcov passes arguments that rq.fit doesn't like, preventing the
command from executing. Here is an example:

 e<-bootcov(rq(y~x),clust,B=10,fitter=rq.fit)

(where clust is my clustering variable) results in

Error in rq.fit.br(x, y, tau = tau, ...) :
  unused argument(s) (maxit = 15, penalty.matrix = NULL)

In contrast, the lm.fit function seems to just ignore these arguments,
resulting in the following warning:

10: In fitter(X[obs, , drop = FALSE], Y[obs, , drop = FALSE], maxit = maxit,  :
  extra arguments maxitpenalty.matrix are just disregarded.

Is there a way that I can either (a) modify bootcov so that it doesn't
pass these arguments or (b) modify rq so that it ignores them?

Thanks in advance,
John Gardner

Quantile regression is now supported in the Design package and bootcov 
has been updated accordingly.  To get the new code run

require(Design)
source('http://biostat.mc.vanderbilt.edu/cgi-bin/viewvc.cgi/*checkout*/Design/trunk/R/bootcov.s')
source('http://biostat.mc.vanderbilt.edu/cgi-bin/viewvc.cgi/*checkout*/Design/trunk/R/Rq.s')
f <- Rq(y ~ rcs(age,4)*sex, x=TRUE, y=TRUE, tau=.25)
b <- bootcov(f, B=1000, pr=TRUE)
nomogram(f)
anova(b)
summary(b, age=c(21,65), sex='male')

This allows one to do quantile regression (but only one tau at a time) 
and get effect plots, nomograms, anova, general contrasts, etc.

Frank


John Gardner wrote:
> I have a quick question, and I apologize in advance if, in asking, I
> expose my woeful ignorance of R and its packages. I am trying to use
> the bootcov function to estimate the standard errors for some
> regression quantiles using a cluster bootstrap. However, it seems that
> bootcov passes arguments that rq.fit doesn't like, preventing the
> command from executing. Here is an example:
> 
>  e<-bootcov(rq(y~x),clust,B=10,fitter=rq.fit)
> 
> (where clust is my clustering variable) results in
> 
> Error in rq.fit.br(x, y, tau = tau, ...) :
>   unused argument(s) (maxit = 15, penalty.matrix = NULL)
> 
> In contrast, the lm.fit function seems to just ignore these arguments,
> resulting in the following warning:
> 
> 10: In fitter(X[obs, , drop = FALSE], Y[obs, , drop = FALSE], maxit =
maxit,  :
>   extra arguments maxitpenalty.matrix are just disregarded.
> 
> Is there a way that I can either (a) modify bootcov so that it doesn't
> pass these arguments or (b) modify rq so that it ignores them?
> 
> Thanks in advance,
> John Gardner
> 
> ______________________________________________
> 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.
> 

-- 
Frank E Harrell Jr   Professor and Chair           School of Medicine
                      Department of Biostatistics   Vanderbilt University