RE: Find a dood function for fitting data
- To: mathgroup at smc.vnet.net
- Subject: [mg42131] RE: [mg42096] Find a dood function for fitting data
- From: "David Park" <djmp at earthlink.net>
- Date: Fri, 20 Jun 2003 04:57:30 -0400 (EDT)
- Sender: owner-wri-mathgroup at wolfram.com
Guillermo,
What do you mean by a good fit? If the data are supposed to follow a
definite law, and they don't, then there is simply a lot of error in the
data. If you are trying to find a law that the data should fit, then it is
up to you to fish around for a form that gives a good fit, perhaps using
information about the process you are dealing with. Did Max Planck write to
something like MathGroup complaining about the poor fit to the black body
spectrum data?
I you just want a continuous curve that goes through all the points, why not
use interpolation?
data = {{0.5, 0.089}, {0.7, 0.12}, {1, 0.1652},
{2, 0.254}, {3, 0.3}, {4, 0.325}, {5, 0.34},
{7, 0.35}, {10, 0.35}, {15, 0.33}, {20, 0.32}};
f[t_] = Interpolation[data, InterpolationOrder -> 1][t]
InterpolatingFunction[{{0.5, 20.}}, "<>"][t]
Plot[f[x1], {x1, 0.5, 20}, Epilog ->
{AbsolutePointSize[4], Point /@ data},
PlotRange -> All];
David Park
djmp at earthlink.net
http://home.earthlink.net/~djmp/
From: guillerm at usal.es [mailto:guillerm at usal.es]
To: mathgroup at smc.vnet.net
(*Dear friend, I want fit these data *)
data = {{0.5, 0.089}, {0.7, 0.12}, {1, 0.1652}, {2, 0.254}, {3, 0.3}, {
4, 0.325}, {5, 0.34}, {7, 0.35}, {10, 0.35}, {15, 0.33}, {20,
0.32}};
(*I have tested with a few function such as that the follow, but I
have not found a good fit*)
Needs["Statistics`NonlinearFit`"];
fit[x1_] = NonlinearFit[data, a1 x1^b1 Exp[b2 x1], {x1}, {a1, b1, b2}]
Plot[fit[x1], {x1, 0.5, 20}, Epilog -> Map[Point, data]]
(*I have the same problem with this other data*)
data1= {{0.5, 0.11}, {0.7, 0.15}, {1, 0.2112}, {2, 0.322}, {3, 0.37}, {4,
0.391}, {5, 0.399}, {7, 0.4}, {10, 0.38}, {15, 0.36}, {20, 0.33}};
(*I will appreciate if you find a good fit, with no many parameters, for
both
data. Thanks
Guillermo
Sanchez*)
---------------------------------------------
This message was sent using Endymion MailMan.
http://www.endymion.com/products/mailman/