I am currently helping someone to fit a logistic regression to a rather
big data set (about 15'000 cases and a 100 variables). Often, the default
number of iterations (10) is not enough to get convergence in the glm
algorithm. Thus we do:
glm.control (maxit = 50)
our.fit <- glm (Y ~ (lengthy list of variables), family = binomial)
This allows us to get a fit. Now we would like to use step.glm to reduce
the variables, but step.glm does maximally 10 iterations and not the
number of iterations given with maxit. Is there any way to force step.glm
to increase the number of iterations?
I hope I have been specific enough. Thanks for thoughts and help!
Regards,
-- Lorenz Gygax
LL GGGGGG Lorenz Gygax room: 36-L-40 LL GG GG Department of Applied Mathematics LL GG G University of Zuerich-Irchel LL GG Winterthurerstr. 190 LL GG GGGG CH-8057 Zurich / Switzerland LL GG GG voice: 41-1-635-58-52 fax: 41-1-635-57-05 LLLLLLL GGGGGGG e-mail: lgygax@amath.unizh.ch
privat: Dennlerstr. 23, CH-8047 Zuerich, voice: 41-1-493-57-05
----------------------------------------------------------------------- This message was distributed by s-news@wubios.wustl.edu. To unsubscribe send e-mail to s-news-request@wubios.wustl.edu with the BODY of the message: unsubscribe s-news