Mathematica 9 is now available
Services & Resources / Wolfram Forums / MathGroup Archive
-----

MathGroup Archive 2001

[Date Index] [Thread Index] [Author Index]

Search the Archive

NonlinearFit work but NonlinearRegress, why?

  • To: mathgroup at smc.vnet.net
  • Subject: [mg29086] NonlinearFit work but NonlinearRegress, why?
  • From: "J. Guillermo Sanchez" <guillerm at gugu.usal.es>
  • Date: Sun, 27 May 2001 18:04:52 -0400 (EDT)
  • Sender: owner-wri-mathgroup at wolfram.com

I wish fit a13, a21, a31 with the experimental values given by "data" as 
follow:

A = {{-0.15` - a13, a21, a31}, {0.15`, -0.17` - a21, 0}, {a13, 0, -a31}};


model3[t_] := MatrixExp[A t].{0.7 Exp[-0.1 t], 0, 0};


data = {{0.`, 0.`}, {1.`, 0.073`}, {2.`, 0.099`}, {3.`, 0.1`}, {4.`,
        0.088`}, {5.`, 0.077`}, {6.`, 0.064`}, {7.`, 0.056`}, {8.`,
        0.043`}, {9.`, 0.039`}, {10.`, 0.028`}};

(*I use NonlinearFit using that start points the same that I have obtained
using FindMinimum {a21 -> 0.1, a13 -> 0.2, a31 -> 0.04}*)

Needs["Statistics`NonlinearFit`"]

(* data should be fitted to second output of model3[t], I mean 
model3[t][[2]]*)


NonlinearFit[data, (model3[t] /. {a21 -> A21, a13 -> A13, a31 -> 
A31})[[2]], {t}, {{A21, 0.1}, {A13, 0.2}, {A31, 0.04}}]

It is Ok, but I need to know  A21, A13, A31 values and their interval of 
conficence, It should be posible using NonlinearRegress but it dont work

NonlinearRegress[data, (model3[t] /. {a21 -> A21, a13 -> A13, a31 -> 
A31})[[2]], {t}, {{A21, 0.1}, {A13, 0.2}, {A31, 0.04}}]

Any help?


  • Prev by Date: Re: Forcing numbers to be real.
  • Next by Date: Re: Forcing numbers to be real.
  • Previous by thread: Re: OOP
  • Next by thread: Re: Why can't Nsolve find a solution to this ?