| T O P I C R E V I E W |
| a.abc.b35 |
Posted - 05/18/2011 : 1:12:39 PM
Origin Ver. and Service Release (Select Help-->About Origin): 8 ,SR0
Operating System: windows 7
.............
I have a data set where i do 2 gauss fits (from multi fit). Now sometimes I can see that there comes a pop up message " Chai square is reduced" . You hit ok, it is gone, and if you look at the fit, sometimes for the same data, the fitted curves will have wired shape. The sum will go through the data points, but the peak 1 (say) may be negative (downward) and peak 2 values larger than the actual data ! If I look at the fit report, I find it no different than the usual case when that pop up window does not appear. Again, sometimes, the fit is OK (both peak1 and peak2 are reasonable and lower than the data, i.e, the lie inside the data and the sum goes through the data )even with that pop up. And if I again look at thee fit report (especially red. chai sqaure) , there is hardly any difference in the values.
Q.1) So can anyone tell me when that pop up window shows up ? In other words, what shall I think about my fit if i see that pop up about chai square ?
Q.2) Is there a way I can see the residual plots in the multi fit analysis (as can be seen if one does single peak fit from NLFIT dialog) ?
AB |
| 3 L A T E S T R E P L I E S (Newest First) |
| jibinene |
Posted - 05/23/2011 : 12:30:29 AM
hi friend u have good tutorial regarding this u will find it in data validtion |
| easwar |
Posted - 05/19/2011 : 4:27:46 PM
Hi AB,
From version 8.0 to 8.1. the multiple peak fit tool did not produce an NLFit report. Instead it created a simplified report, so when one tries "change parameter" to further control the fit, there is no option to set bounds or fix some parameter etc.
We have resolved this in the latest version:
http://originlab.com/index.aspx?go=Products/Origin/DataAnalysis/PeakAnalysis/PeakFitting&pid=1760
So in its new form in 8.5.1, you can click and select as many peaks as you decide, then you have option to fit, or open NLFit immediately to place some bounds or constraints, or share parameters etc. You can also first fit and then re-open the NLFit dialog using Change Parameters menu on the lock, to further control the fit.
Often with data where peaks may not be clearly defined/separate, it is possible the iterative procedure may flip the direction of one peak to get a better fit for the overall curve, and only way to prevent that is to put say bounds of A>0 for all peaks.
So please try with the 8.5.1 demo to see if that works better.
If you need further help send your data to tech support.
In 8.5.1 there is also a Quick Peaks gadget which may be of interest to you:
http://originlab.com/index.aspx?go=Products/Origin/DataAnalysis/Gadgets/QuickPeaks
Easwar
OriginLab
|
| a.abc.b35 |
Posted - 05/19/2011 : 3:17:20 PM
No takers...huh ?
AB |
|
|
