R help - Jul 2003 - Nonliner Rgression using Neural Nnetworks

Hi, 
I am an old hand at chemistry but a complete beginner at statistics
 including R computations.
My question is whether you can carry out nonlinear
multivariate regression  analysis in  R using neural networks, where the
output variable can range from -Inf to  + Inf., unlike discriminant 
analysis where the output is confined to one  or zero. The library nnet
seems to work only in the latter case but then I could  be wrong. 

Please help me there.

Thanks in advance.

Y.Ishii <yukiasais at ybb.ne.jp>
2-3-28?Tsurumaki-minami, Hadano
257-0002 Japan

On Fri, 11 Jul 2003 18:56:58 +0900
Yukihiro Ishii <fwhd3550 at mb.infoweb.ne.jp> wrote:
> Hi, 
> I am an old hand at chemistry but a complete beginner at statistics
>  including R computations.
> My question is whether you can carry out nonlinear
> multivariate regression  analysis in  R using neural networks, where the
> output variable can range from -Inf to  + Inf., unlike discriminant 
> analysis where the output is confined to one  or zero. The library nnet
> seems to work only in the latter case but then I could  be wrong. 
> 
> Please help me there.
> 
> Thanks in advance.
> 
> Y.Ishii <yukiasais at ybb.ne.jp>
> 257-0002 Japan
You might want to look at the paper at

http://brain.cs.unr.edu/publications/goodman.ann_advantages.jasa99.pdf

The work was done using a nice standalone neural net program Nevprop by Goodman
and colleagues, which is intended for binary outcomes and incorporates
bootstrapping for estimating predictive accuracy of the network.

You may obtain Nevprop at http://brain.cs.unr.edu
---
Frank E Harrell Jr              Prof. of Biostatistics & Statistics
Div. of Biostatistics & Epidem. Dept. of Health Evaluation Sciences
U. Virginia School of Medicine  http://hesweb1.med.virginia.edu/biostat

On 11 Jul 2003 at 18:56, Yukihiro Ishii wrote:
> Hi, 
> I am an old hand at chemistry but a complete beginner at statistics
>  including R computations.
> My question is whether you can carry out nonlinear
> multivariate regression  analysis in  R using neural networks, where the
> output variable can range from -Inf to  + Inf., unlike discriminant 
> analysis where the output is confined to one  or zero. The library nnet
> seems to work only in the latter case but then I could  be wrong. 
You are wrong. nnet can be used to predict a continous variable, for 
instance  by setting the arguments linout=TRUE. 

For ways to set different types of networks, see 
?nnet

ans especially the arguments
linout
entropy
softmax
censored

Kjetil Halvorsen
> 
> Please help me there.
> 
> Thanks in advance.
> 
> Y.Ishii <yukiasais at ybb.ne.jp>
> 2-3-28 $B!! (BTsurumaki-minami, Hadano
> 257-0002 Japan
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-help