This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: [Help-gsl] robust regression with GSL ?
- From: Brian Gough <bjg at network-theory dot co dot uk>
- To: Axel Kowald <kowald at molgen dot mpg dot de>
- Cc: help-gsl at gnu dot org, gsl-discuss at sources dot redhat dot com
- Date: Wed, 12 Nov 2003 13:32:28 +0000
- Subject: Re: [Help-gsl] robust regression with GSL ?
- References: <5.2.0.9.2.20031110105046.00acf900@harry.molgen.mpg.de>
Axel Kowald writes:
> Hello everybody,
>
> I just discovered GSL and wonder if there are routines for doing robust
> linear regression with it (at first sight I didn't find anything) ?
> I read about robust regression that uses M-estimators with Andrew's Sine,
> Huber's method or Tukey's Biweight as influence function.
>
> Exists such a routine in GSL ?
No, although the existing least squares routines could probably be
adapted to use a different metric.
> And if not, does anybody now from where to get C/C++ source code for robust
> regression ?
>
This is something it would be useful to have in GSL.
--
Brian Gough