R-beta: port of bicreg package to R?

Thomas Lumley thomas at biostat.washington.edu
Thu Sep 24 19:03:59 CEST 1998

On Thu, 24 Sep 1998, Malcolm Gillies wrote:
> I still don't have any joy, as now leaps is complaining. This probably
> has more to do with the fact that I'm flying blind...
> > bicreg(hiv[-14,-17], hiv[-14,17])
> Error: leaps requires full-rank design matrix; use subsets()
> > traceback()
> [1] "stop(\"leaps requires full-rank design matrix; use subsets()\")"
> [2] "leaps(x, y, wt = wt, method = \"r2\", names = dimnames(x)[[2]])"
> [3] "bicreg(hiv[-14, -17], hiv[-14, 17])"                            

*Is* your design matrix singular? -- try running lsfit() on it. If it is
then leaps() won't work (though the Fortran code is still ok).  More
importantly, I think Occam's Window is an asymptotic technique -- if you
have a singular design matrix it probably isn't reliable.

So, first try using 
R> lsfit(hiv[-14, -17], hiv[-14, 17])
 to see if the problem is with the data. Then use d
R> debug(bicreg)
R> bicreg(hiv[-14, -17], hiv[-14, 17])
and step through the function until the leaps() line, then do
Browse> lsfit(x,y,wt=wt)
to see if the design matrix is still non-singular.  If it is, there is a
bug in leaps; otherwise the design matrix has been mangled in some way


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