# Re: Why this is so difficult to do curve fitting?

**From:** Uwe Schmitt (*schmitt_at_num.uni-sb.de*)

**Date:** 08/16/04

**Next message:**lucy: "Re: how to decide a good sampling rate for sampling a function without obvious frequency?"**Previous message:**Uwe Schmitt: "approximating a matrix"**In reply to:**ip4ram_at_yahoo.com: "Re: Why this is so difficult to do curve fitting?"**Next in thread:**Martin Brown: "Re: Why this is so difficult to do curve fitting?"**Messages sorted by:**[ date ] [ thread ]

Date: Mon, 16 Aug 2004 17:42:07 +0200

*>
*

*> take a look at polynomial interpolation.The solution might not be of
*

*> the form y = a*(x/255).^b+ c ,but might fit the data more accurately
*

*>
*

*>
*

*> http://mathworld.wolfram.com/LagrangeInterpolatingPolynomial.html
*

*>
*

*> ram
*

I do not think that this is a good idea. Interpolating the given

data requires a polynomial of degree seven. Interpolating gets

very unstable for degrees greater four, that is: minimal change

of the data leads to big derivations in the coefficients of

the fitting polynomial and thus big derivations if evaluating

this polynomial.

Especially as lucy asks for fitting, I assume that the given

data are noisy and that it is not assumed that the searched for

function exactly hits the given 2d points.

Greetings, Uwe.

**Next message:**lucy: "Re: how to decide a good sampling rate for sampling a function without obvious frequency?"**Previous message:**Uwe Schmitt: "approximating a matrix"**In reply to:**ip4ram_at_yahoo.com: "Re: Why this is so difficult to do curve fitting?"**Next in thread:**Martin Brown: "Re: Why this is so difficult to do curve fitting?"**Messages sorted by:**[ date ] [ thread ]