| Author |
 Topic  |
|
|
mej
Albania
Posts |
 Posted - 10/05/2004 : 10:03:40 AM
|
Hi everybody
I am trying to use origin 6.1 Pro to do non-linear fit on some data, but I want to consider the error bar for each data in my fitting. My
error bar is about 10 % for each data
Thanks |
|
|
easwar
USA
1965 Posts |
Posted - 10/05/2004 : 1:09:40 PM
|
Hi,
The fitter supports fitting with error bars. While in the NLSF Advanced fitter dialog, go to the Options->Control panel and you will see a drop-down for "Weighting method". Different methods such as instrumental, statistical, direct etc. are available and these are documented in the help files.
Do you have error bar columns associated with your data? Are you asking about creating error bar columns?
Easwar
OriginLab
|
 |
|
|
mej
Albania
Posts |
Posted - 10/13/2004 : 12:30:44 PM
|
Yes I have error bar columns associated with your data
|
|
|
|
vhable
Germany
Posts |
Posted - 10/14/2004 : 05:53:22 AM
|
Hello!
I got a question in a similar direction: When I do a gauss-fit, I get for the "w" a value plusminus an error. How will this error be calculated? Are error bars in my data always considered in the calculation of this error or only when I choose the "instrumental weighting method"?
Thanks and best regards from Germany! |
|
|
|
easwar
USA
1965 Posts |
Posted - 10/14/2004 : 09:50:56 AM
|
mej: If you have error bar associated with the data, you can first make a plot by selecting both the Y and YErr columns. Then bring up the NLSF tool. You will still need to go to the Options->Control menu in the NLSF tool and then pick a weighting method. With error bar data, the typical weighting method used is Instrumental. To learn about the various weighting options, press the F1 help key while on that dialog page to bring up the help file section.
vhable: Please see reply above to mej on how to get information on weighting methods.
Now, the error computed on each fitting parameter such as w is the standard error. The help file documents how this is computed from the minimization procedure. This error parameter is computed irrespective of whether your data had associated error bars/weights or not.
Easwar
OriginLab
|
|
|
|
vhable
Germany
Posts |
Posted - 10/14/2004 : 10:08:16 AM
|
Thank you for your information!
Many greetings,
Volker |
|
|
| |
Topic |
|
error bar and non-linear fitting