R help - Sep 2011 - Constrained regressions (suggestions welcome)

All,

Could anyone recommend a package that allows the user to constrain the
coefficients from a multiple regression equation?  

I tried using the  gl1ce function in lasso2, but couldn't get it to
work.  I created a contrived example to illustrate my starting point.

data(cars)
 fmla <- formula(dist ~ speed)
 gl1c.E <- gl1ce(fmla, data = cars)
 gl1c.E
 gl1c.E <- gl1ce(fmla, data = cars, link=identity())
 gl1c.E

I started with the example, but I think my data is more Gaussian than
binomial.  This returns the following errors.
>      data(cars)
>      fmla <- formula(dist ~ speed)
>      gl1c.E <- gl1ce(fmla, data = cars)
Error in family(family) : 
  link "family" not available for gaussian family; available links are
'inverse', 'log', 'identity'
> source(.trPaths[5], echo=TRUE, max.deparse.length=150)
>      gl1c.E <- gl1ce(fmla, data = cars, link=identity())
Error in identity() : argument "x" is missing, with no default

In this example I haven't even gotten to the stage of constraining the
coefficients yet because I can't get the function to work with minimal
inputs.

Thanks,

RogerRoger J. Bos, CFA
Rothschild Asset Management, Inc.
Tel     +1 (212) 405-5471
Email roger.bos at rothschild.com
1251 Avenue of the Americas, NY, NY  10020
***************************************************************

This message is for the named person's use only. It may\...{{dropped:14}}

Roger:

1. Constrain how? Via a penalty? Via a linear or nonlinear equations or
inequalities on the coefficients? You need to be explicit.

2. The above point and:

I started with the example, but I think my data is more Gaussian
than
> binomial.  This returns the following errors.
> --
>
both indicate to me that your math/statistical skills may be inadequate for
the task. As a result, I think you would do well to consult a local
statistician -- or someone with adequate statistical background in this
area,anyway. Good people on this list may offer advice, but you may be
heading down the wrong path and thereby committing errors of the third kind
(right answer to wrong question).

Cheers,
Bert




"Men by nature long to get on to the ultimate truths, and will often be
impatient with elementary studies or fight shy of them. If it were possible
to reach the ultimate truths without the elementary studies usually prefixed
to them, these would not be preparatory studies but superfluous
diversions."

-- Maimonides (1135-1204)

Bert Gunter
Genentech Nonclinical Biostatistics
467-7374
http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm

	[[alternative HTML version deleted]]