[R] Error messages

[Prof Brian D Ripley]

On 4 Jan 1999, Peter Dalgaard BSA wrote:

> [I seem to recall that there's a reason why lm.fit sticks in NA
> (rather than zero) for undetermined coefficients, but I can't offhand
> remember it. I suppose that lm.influence() would make sense even for
> singular designs, so the current behaviour is likely a bug, not a
> feature]

Yes, that's how you determine that the coefficients are not determined!
It is the simplest way to find out exactly how the degrees of freedom are
allocated to terms in models with factors, and also helps in omitting
values in printouts (rather than showing them as zero).

As for the bug, I think putting in zero would be a bug too, as the qr
matrix gets pivoted for singular models, and I see no sign of the pivots
being passed down to the Fortran program. At least lm.influence
should report on singular models (the S version handles them).

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._