Oleg recently posted a link for J and (the
statistical package) R. Using that link I found a
description of plain logistic regression at
http://zoonek2.free.fr/UNIX/48_R/12.html#4 . A stepwise
version means selecting predictor variables one (or more) at
a time, based on the improvement in the resulting R^2 (index
of determination). According to the link supplied by Oleg, a
stepwise version was not available in R.
A slightly more complete description of logistic
regression is given at
http://www.statsoftinc.com/textbook/stathome.html . In the
table of contents in the righthand pane click on
"Generalized Linear Mod." ("GLZ").
The description of plain logistic regression is
likely enough to produce a J version. Newton-Raphson
iteration method is commonly used for solving the maximum
likelihood equation for the desired paramter estimates.
(B=)
Brian Schott
Atlanta, GA, USA
schott DOT bee are eye eh en AT gee em ae eye el DOT com
http://www.cpu-net.com/host/bschott/
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
