| Author |
 Topic  |
|
|
pedro.sanches
Netherlands
2 Posts |
 Posted - 11/16/2012 : 05:09:24 AM
|
Origin Ver. and Service Release (Select Help-->About Origin): OriginPro 9.00 (64-bit) b45 (evaluation)
Operating System: Windows 7 64 bit
Dear Originlab,
I installed the evaluation version to try out the Fit Compare options.
I have two data sets fitted with an ExpDec1 function (ungrouped mode).
I tried to compare using Fit compare (Analysis -> Fitting...->Compare Data sets)
Recalculate is set to Auto, I choose the corresponding Fit Sheets, and run.
The F-test result however says "There is not enough information to draw conclusion"
Is there something I could do about it? Would I need to have more points? I mean both curves fitted with R^2 > 0.87...
Thank you for your help!
Best Regards,
Pedro Sanches
|
|
|
Sam Fang
293 Posts |
Posted - 11/19/2012 : 04:59:36 AM
|
Did you select Analysis: Fitting: Nonlinear Curve Fit or Analysis: Fitting: Exponential Fit from Origin menu for fitting? You'd better use Nonlinear Curve Fit dialog. It is a known bug. It has been fixed in our inhouse build.
Is your F-value negative or a missing value in the F-test table? If it is negative, you have to calculate the correct Prob>F value manually. Run the following script in the command window, e.g. in your F-test table, F=-1e-3, Numer.DF=3, Denom.DF=194, then the correct Prob>F value should be:
1-fcdf(1e-3,3,194)=
And the P-value should be 1. Because P-value>0.05, it can be concluded that they are not different. It is also a known bug and fixed in our inhouse build. It sometimes occurs when the two datasets are completely same, and the negative value is due to the accuracy error.
If your case does not attribute to the above two reasons, you can send your project to us. Just click Send File to Tech support button in the top right of the forum.
Sam
OriginLab Technical Services |
Edited by - Sam Fang on 11/19/2012 05:08:56 AM |
 |
|
|
pedro.sanches
Netherlands
2 Posts |
Posted - 11/20/2012 : 07:44:16 AM
|
Dear Sam,
Thanks for the info!
I had used the Nonlinear Curve fit and the F-value was indeed negative.
I have run the recommended script with the Compare Datasets results (-0.32906,3,4) and I have obtained a p-value = 0,8059. Thus, no statistical significance difference found.
Best Regards,
Pedro |
|
|
| |
Topic |
|
|
|
Compare Data sets