From mboxrd@z Thu Jan  1 00:00:00 1970
From: Brian Gough <bjg@network-theory.co.uk>
To: "Tom Browder" <tom2@fwb.asi.srs.com>
Cc: <gsl-discuss@sources.redhat.com>
Subject: Re: Mutivariate Regression (GSL Topic: "Fitting")
Date: Wed, 19 Dec 2001 13:20:00 -0000
Message-id: <15201.30234.816726.23366@debian>
References: <01c1169d$40b58900$cb01000a@tombrowd>
X-SW-Source: 2001/msg00334.html

Tom Browder writes:
 > I just started using GSL and wonder if it is possible to allow for
 > a constant in the multifit function?

It should be possible. For gsl_multifit_linear use a column of 1's in
the matrix 'X'.  For the non-linear case it should work the same as
other parameters.  If it does not work let us know.

 >  The following would also be nice additions to the multifit
 > function suite (see MS Excel regression output for example):
 >  1. Produce the correlation coefficient (r) and other statistics.
 > 2. Allow fit variable weighting (not observation weighting).
 > 3. Allow for principal factor computations.

Thanks for the suggestions, I'll add them to the TODO file in case
anyone wants to work on them.

regards
Brian Gough