Skip to content
Prev 8589 / 398502 Next

sm library

On Tue, 19 Dec 2000, Oscar wrote:

            
Please look in the archives, as I posted the solution a day or so ago.
The bug is that glm.fit has been broken since that function was ported to
R, and now gives linear predictor 0 for weight-zero cases, incorrectly.

The best solution is to get the new sm2 package, on its way to CRAN now.
You can pick up a version from
http://www.stats.ox.ac.uk/pub/R/sm_2.0-1.tar.gz or a Windows port at
http://www.stats.ox.ac.uk/pub/RWin/sm.zip (for rw1020), both as from about
five minutes ago.