The Origin Forum
File Exchange
Try Origin for Free
The Origin Forum
Home | Profile | Register | Active Topics | Members | Search | FAQ | Send File to Tech support
Username:
Password:
Save Password
Forgot your Password? | Admin Options

 All Forums
 Origin Forum
 Origin Forum
 Exponential fit statistics?
 New Topic  Reply to Topic
 Printer Friendly
Author Previous Topic Topic Next Topic Lock Topic Edit Topic Delete Topic New Topic Reply to Topic

hdvorak

Pasadena, CA
4 Posts

Posted - 02/14/1999 :  6:33:00 PM  Show Profile  Edit Topic  Reply with Quote  View user's IP address  Delete Topic
Is there any way to get statistics on how well data are fit by a first-order exponential decay curve? When I use Analysis/Fit Exponential Decay/First Order I get parameters and errors on the parameters, but I was wondering if there's something analagous to R^2 that can be produced for a fit to an exponential. It seems like there may be some way to do it through the Advanced form of the Non-Linear Curve Fit tool, but I was totally overwhelmed by the description in the manual -- it seemed like a lot of the stuff there was overkill for what I want to do... Any suggestions?

rtoomey

USA
184 Posts

Posted - 02/16/1999 :  2:02:00 PM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply


Response to Exponential Fit statistics?


Fitting routines such as Analysis => Fit => Exponential Decay run the nonlinear curve fitter in the background. Since the fitter is involved in the process, there are several results that can be obtained. To get these results, do the following:



  1. Perform the fit from the menu.

  2. Select Analysis => Nonlinear Curve Fit.

  3. Click the MORE button (if it is available) to take you to the advanced mode of the fitter.

  4. Select Action => Results from within the fitter.

  5. Click the Parameters Worksheet button.


At this point, a worksheet should be created that contains several different values, one of which is the COD (also known as R^2). If you are interested in finding out what the rest of these values mean, I have compiled a listing of pages in the Origin 5.0 LabTalk and User's Manuals (English) to go to for more information.


If you feel that the information this supplies you with is not enough, the only suggestion I can offer you would be to consult our list of references (page 601 of the Origin 5.0 User's Manual).


The List:


************


  • SSR - The value of the sum of squares function - p.270 LabTalk(nlsf.ssr), p. 363-364 (buil-in function reference discussion of ss(p) function)

  • COD(R^2) - The coefficient of determination - pp.469-475 Origin User's Manual (theory), p.266 LabTalk (nlsf.cod)

  • MuFinal - The resultant value of the diagonal magnification of the curvature matrix calculated during fitting - p.269 LabTalk (related object properties are: nlsf.Mu and nlsf.MuMin)

  • DerivStep - The amount parameters are allowed to vary during(numeric calculation of derivative - during fitting) p.267 LabTalk (nlsf.DerivStep)

  • ParamChange - Number of Parameters that are allowed to vary - p.269 LabTalk (nlsf.nParaVary)

  • Constr - The number of constraints you imposed - p.269 LabTalk (nlsf.nConstr)

  • ConstrEff - The number of effective constraints - p.269 LabTalk(nlsf.nConstrEff)


In general, the major areas of the manuals to concentrate on when it comes to fitting theory and parameters would be:



  • LabTalk pp. 266-278 (the nlsf object properties and methods)

  • 5.0 Users pp. 462-475 (fitting theory)


************


The Parameters Worksheet is just one of the many features of the fitter's Results dialog box that allows you to analyze how well the fit has taken place. You may also be interested in the Prediction and Confidence Bands, as well as the Residue plot. The Residue plot (a plot of the residual values found when subtracting the fit data from your raw data) is particularly useful when determining how closely a function has fitted your data.


I hope that this information proves helpful.


Sincerely,

Ryan Toomey



[This message has been edited by rtoomey (edited 02-16-99).]

Go to Top of Page

hdvorak

Pasadena, CA
4 Posts

Posted - 02/16/1999 :  6:02:00 PM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
That's exactly what I was looking for. I knew it had to be in there somewhere... Thank you very much!
Go to Top of Page
  Previous Topic Topic Next Topic Lock Topic Edit Topic Delete Topic New Topic Reply to Topic
 New Topic  Reply to Topic
 Printer Friendly
Jump To:
The Origin Forum © 2020 Originlab Corporation Go To Top Of Page
Snitz Forums 2000