| Author |
 Topic  |
|
|
peter.cook
UK
356 Posts |
 Posted - 05/24/2005 : 08:40:51 AM
|
Origin Version (Select Help-->About Origin): Origin 7.5 SR5
Operating System: Win 2000
Hi,
I'd be interested in knowing some detail / opinions about the differences between a single multiple fit vs a multiple single fit.
i) When should either be used or not used?
ii) What are the essential differences?
iii) I assume global fitting is equvalent to a multiple fit
iv) Can I always use a single multiple fit in my code ie avoid having two separate bits of code. If this is so, how can I pull out eg individual R and nlsf.ssr values for each dataset. Parameter values and standard errors are different of course.
Cheers,
Pete
|
|
|
easwar
USA
1964 Posts |
Posted - 05/24/2005 : 09:53:28 AM
|
Hi Pete,
If you perform a single fit with multiple datasets, the fit parameters corresponding to all datasets are combined and the minimization occurs in this combined parameter space. Thus only one value will be generated for fit statistics such as SSR, R-Srq etc which then represent global fit statistics. Thus this method is best suited when you really want to share parameters while fitting - such as say share the centroid parameter when fitting 5 gaussian-like datasets with the gaussian function, where you know/decide that the centroids should be the same for each fit.
On the other hand if you are interested in fitting each dataset with the fitting function and generate goodness of fit statistics separately for each dataset, then you should do multiple single fits.
Easwar
OriginLab
|
 |
|
|
peter.cook
UK
356 Posts |
Posted - 05/24/2005 : 3:26:23 PM
|
Hi Easwar,
Makes sense thanks. Just wondered if I was missing a trick.
Cheers,
Pete
|
|
|
|
peter.cook
UK
356 Posts |
Posted - 06/09/2005 : 06:57:47 AM
|
Hi Easwar,
Presumably you could still obtain statistical values for each dataset after a global fit eg the equivalent of nlsf.iterate(0) for each dataset with the global parameters. Is this worth considering as a feture?
Cheers,
Pete
|
|
|
| |
Topic |
|
|
|
Multiple fits vs single multiple fit