I have about 20 000 cases with discrete variables (some are counts,
some are factors). I'm interested in fitting a series of models
outcome ~ 1
outcome ~ sex
outcome ~ sex + age
outcome ~ age * sex
outcome ~ age * sex + location
...
I do NOT expect to get any statistical significance out of this
Perhaps step(lm.object)
helps, what use a backward elimination of attributes ,
further i remind in library(leaps) some helpfuel things
for modellimg selection.
christian
Am Dienstag, 16. März 2004 05:53 schrieb Richard A. O'Keefe:
I have about 20 000 cases with discrete variables (some