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
 NLSF & fit comparison		  New Topic  Reply to Topic
 Printer Friendly
Author Previous Topic Topic Next Topic Lock Topic Edit Topic Delete Topic New Topic Reply to Topic

Lemming76

Austria
Posts
Posted - 04/18/2007 :  08:29:41 AM  Show Profile  Edit Topic  Reply with Quote  View user's IP address  Delete Topic
Origin Version: OroginPro 7.5G SR6 v7.5886 (B886)
Operating System: WinXP SP2

Hi!
I have 2 questions concerning the nonlinear fitting functionalities:

1. I used the NLSF wizard to fit curves to multiple datasets using separate parameter sets. I noticed that the errors reported for the parameters are different from when I fit each dataset separately. Is there any (statistical) sense behind this?

2. the fit comparison tool reports the F- and P-values for the F-test. The Origin reference states that F=((SSRcomb-SSRsep)/(DOFcomb-DOFsep))/(SSRsep/DOFsep) and according to what I've been reading elsewhere this should be the correct formula.
However, the tool actually calculates according to the formula F=((SSRcomb-SSRsep)/(DOFcomb-DOFsep))/SSRsep resulting in false test conclusions. Did I miss something or is this a software bug?

thanks for any replies!
Chris

larry_lan

China
Posts
Posted - 04/18/2007 :  10:53:44 PM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hi Chris:

1. This is because if you fit multiple datasets in one time, Origin actually performs "Global Fit". That is, it will treat all inputs as one curve, and generates one reduced Chi^2, DOF, etc. And the parameters error related to such values, so the result is different.

2. Yes, this is a bug. And we have fixed it in our next version.

Thanks

Larry
OriginLab Technical Services
Go to Top of Page

Wendy_Yang

USA
Posts
Posted - 04/30/2009 :  1:25:24 PM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hi Larry, this is a quesiton related to your answer #1. Is it possible to obtain the Chi^2 value for each dataset when performing global fit? The result only contains the chi^2/DoF for all the datasets, but I would like to know how well the individual curves are fitted. I think I know how to calculate chi^2 by hand, but it will be very helpful if there is an easier way to get them from the software.

Thanks!
Wendy
Go to Top of Page

Deanna

China
Posts
Posted - 05/04/2009 :  01:57:26 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hi Wendy,

A global fit fits all datasets simultaneously. So there is only one chi square value.

If you do not need to share any parameters, please use the independent fitting mode instead. In this case, each dataset is fitted separately and there will be one chi-square value for each fitting (dataset).

Please read:
http://www.originlab.com/www/helponline/Origin8/en/mergedProjects/QuickHelp/QuickHelp/What_is_the_difference_between_global_fit_and_independent_fit.html

Deanna
OriginLab Technical Services
Go to Top of Page

Wendy_Yang

USA
Posts
Posted - 05/06/2009 :  11:48:59 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hi Deanna,

Thank you very much! I actually need to share all parameter except for one, so the global fit works for me. I decided to do residual plots for the individual datasets, and that seems to help with the visualization.

Thanks again,
Wendy
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