[R] problem with cov.* in package lqs

Liaw, Andy andy_liaw at merck.com
Wed Jan 24 18:01:38 CET 2001

> From: Prof Brian Ripley [mailto:ripley at stats.ox.ac.uk]
> On Wed, 24 Jan 2001, Liaw, Andy wrote:
> > I ran into an error message when I use the cov.* (rob, mcd, 
> mve) function in
> > the package lqs.  The data matrix is 761 rows by 661 
> columns.  I get the
> > following message:
> >
> > 	NA/NaN/Inf in foreign function call (arg 1)
> >
> > Can anyone provide some insight on why this occurs, and 
> better yet, how to
> > get around it?
> Not without more information.  But that is probably in
>         z <- .C("mve_fitlots", as.double(x), as.integer(n), 
> as.integer(p),
> in which case it probably means some column has IQR 0.
> Generally, you need to help yourself on these things.
> 1) traceback() tells you where the error is.
> 2) debug(cov.rob) and re-run the call, and step though a line 
> at a time
> In particular, look at divisor.

Thanks a lot for the pointer.  There are five columns that had IQR=0!

BTW, traceback() didn't give any more information than what I provided.  It
simply said the call bombed inside cov.rob.  I'll have to give debug() a

r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch

More information about the R-help mailing list