[R] Constrained regression

[Mike Cheung]

Dear Carlos,

One approach is to use structural equation modeling (SEM). Some SEM
packages, such as LISREL, Mplus and Mx, allow inequality and nonlinear
constraints. Phantom variables (Rindskopf, 1984) may be used to impose
inequality constraints. Your model is basically:
y = b0 + b1*b1*x1 + b2*b2*x2 +...+ bp*bp*xp + e
1 = b1*b1 + b2*b2 +...+ bp*bp

Alternatively, you can set some condition bounds on the parameter
estimates. Then you only have to impose the second constraint.

Rindskopf, D. (1984). Using phantom and imaginary latent variables to
parameterize constraints in linear structural models. Psychometrika,
49, 37-47.

Regards,
Mike
-- 
---------------------------------------------------------------------
 Mike W.L. Cheung               Phone: (65) 6516-3702
 Department of Psychology       Fax:   (65) 6773-1843
 National University of Singapore
 http://courses.nus.edu.sg/course/psycwlm/internet/
---------------------------------------------------------------------

On Mon, Mar 3, 2008 at 11:52 AM, Carlos Alzola <calzola at cox.net> wrote:
> Dear list members,
>
>  I am trying to get information on how to fit a linear regression with
>  constrained parameters. Specifically, I have 8 predictors , their
>  coeffiecients should all be non-negative and add up to 1. I understand it is
>  a quadratic programming problem but I have no experience in the subject. I
>  searched the archives but the results were inconclusive.
>
>  Could someone provide suggestions and references to the literature, please?
>
>  Thank you very much.
>
>  Carlos
>
>  Carlos Alzola
>  calzola at cox.net
>  (703) 242-6747
>
>
>         [[alternative HTML version deleted]]
>
>  ______________________________________________
>  R-help at 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.
>