One thought is to train the net and obtain a performance measure on a testing corpus. Next, for each input, run the testing corpus again, but zero all values for that input and obtain a measure of performance. Zeroing an important node will hurt performance more than zeroing an unimportant node.
On Tue, Mar 10, 2009 at 9:41 AM, abbas tavassoli <tavassoli...@yahoo.com> wrote: > > Hi, I have a binary variable and many explanatory variables and I want to > use the package "nnet" to model these data, (instead of logistic regression). > I want to find the more effective variables (inputs to the network) in > the neural network model. how can I do this? > thanks. > > > > > [[alternative HTML version deleted]] > > ______________________________________________ > R-help@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. > -- Mike Lawrence Graduate Student Department of Psychology Dalhousie University Looking to arrange a meeting? Check my public calendar: http://tinyurl.com/mikes-public-calendar ~ Certainty is folly... I think. ~ ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.