From help-octave-request at bevo dot che dot wisc dot edu Thu Jan 16 20:10:27 2003
Subject: Re: non linear fit
From: "Henry F. Mollet" <mollet at pacbell dot net>
To: Paolo Ariano <paolo dot ariano at unito dot it>,    Octave_post <help-octave@bevo.che.wisc.edu>
Date: Thu, 16 Jan 2003 18:11:58 -0800

The gnuplot manual (3 Dec. 1998, p. 19 ff) has a statistical overview on
non-linear-least-squares and gives practical guidelines. You have to
A) define your fitting function f(x,y), then
B) fit(x,y) 'datafile' .......
The fitting procedure must use least squares by default and it is probably a
y-on-x non-linear-least square fit. I don't know how you could specify your
own "loss-function", e.g. if you wanted to use max. likelihood instead of
least squares or if you wanted to use a geometrical mean (GM) regression
(would be called for if you have about equal measurement error for both x
and y data points).
Henry

on 1/16/03 7:24 AM, Paolo Ariano at paolo dot ariano at unito dot it wrote:

> hi *,
> at the moment i'm using octave to analyze my data and gnuplot to make
> fit, it is possible also to make non linear fit with octave ?
> 
> thanks a lot
> paolo



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------