R help - Apr 2009 - Constrained, multiple response statistics

R'ers:

    I was hoping I could get some direction on this.  I have a dataset 
of the form:

Y1,Y2,...,YM = f(X1,X2,...,XN), where N is >>> M

The response data (Y1,Y2,...,YM) is frequency data, such that the sum of 
all Yi = 1.0.  Both Xj and Yi are continuous variables.

I'm trying to figure out the best approach(es) to solving for the model 
f() -- any ideas?  I could solve each Y one at a time, but the lack of 
constraint worries me, and I'm pretty sure that normalizing the data 
afterwards to sum to 1.0 is not going to work out properly.  Thoughts?  
I've never worked with multiple response statistics before, so I'm 
mostly trying to get some pointers on where to begin investigating...

--j

-- 

Jonathan A. Greenberg, PhD
Postdoctoral Scholar
Center for Spatial Technologies and Remote Sensing (CSTARS)
University of California, Davis
One Shields Avenue
The Barn, Room 250N
Davis, CA 95616
Cell: 415-794-5043
AIM: jgrn307, MSN: jgrn307 at hotmail.com, Gchat: jgrn307

This sounds very similar to what I've been working on, but I'm not sure
without an example.

My solution has been to use an optimization that normalizes inside the
objective function.  The betas that are provided by optim are not
normalized, however since they were normalized inside the objective
function, normalizing them after the fact mirrors the internal workings of
the objective function.

see this example:
http://markmail.org/message/ze5237m6gbgvvvyf

Still, after looking at several statistical packages, and considering the
thoughtful responses from my post, I think that there must be a better way
using existing models, so I've been looking at other packages / models.

On Tue, Apr 7, 2009 at 6:10 PM, Jonathan Greenberg
<greenberg@ucdavis.edu>wrote:
> R'ers:
>
>   I was hoping I could get some direction on this.  I have a dataset of the
> form:
>
> Y1,Y2,...,YM = f(X1,X2,...,XN), where N is >>> M
>
> The response data (Y1,Y2,...,YM) is frequency data, such that the sum of
> all Yi = 1.0.  Both Xj and Yi are continuous variables.
>
> I'm trying to figure out the best approach(es) to solving for the model
f()
> -- any ideas?  I could solve each Y one at a time, but the lack of
> constraint worries me, and I'm pretty sure that normalizing the data
> afterwards to sum to 1.0 is not going to work out properly.  Thoughts? 
I've
> never worked with multiple response statistics before, so I'm mostly
trying
> to get some pointers on where to begin investigating...
>
> --j
>
> --
>
> Jonathan A. Greenberg, PhD
> Postdoctoral Scholar
> Center for Spatial Technologies and Remote Sensing (CSTARS)
> University of California, Davis
> One Shields Avenue
> The Barn, Room 250N
> Davis, CA 95616
> Cell: 415-794-5043
> AIM: jgrn307, MSN: jgrn307@hotmail.com, Gchat: jgrn307
>
> ______________________________________________
> 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.
>
	[[alternative HTML version deleted]]