| Author |
 Topic  |
|
|
Keq
Germany
4 Posts |
 Posted - 05/19/2016 : 4:27:24 PM
|
hi,
i have some data and need to do a polynomial fit. my problem is that fitted curve has to contain exactly my first data point, but the proposed curves never do that. there is the option to fix intercept at y=.. for example but this way i have to recalculatre the fix y value on my own so that the curve will contain my first data point later. is there any easier possibily to weight data points more than othere or just tell him that the polynom has to start at the first data point?
already thank you for anybody who can help :) |
|
|
meili_yang
103 Posts |
Posted - 05/19/2016 : 4:57:51 PM
|
Hi Keq,
For your case, use NLFit tool instead of Polynomial Fit tool. From NLFit dialog, either to put a linear constraint or use a weighting data column with much bigger value on the first data point, both should work.
Please take a look at following tutorial talking exactly how to do this:
http://www.originlab.com/doc/Tutorials/Fitting-FixThroughPoint
Let me know if still any questions.
Meili
OriginLab Tech Support |
 |
|
|
Keq
Germany
4 Posts |
Posted - 05/19/2016 : 5:02:46 PM
|
| i already tried that one. i added a linear condition sometimes it almost works but most of the time origin tells me "error in conditions-check operators and stuff..." if i redo it several times it works 50%, but origin somehow changes my data instead of changing the fit |
|
|
|
meili_yang
103 Posts |
Posted - 05/20/2016 : 09:33:17 AM
|
Keq,
Maybe to use weighting is better for your case, and you can use Polynomial Fit under Analysis menu since it also supports direct weighting.
So in you data worksheet, add another column for instance with the first value as 100 and rest values as 1 in order to force fitting pass through first point. Then in the Polynomial Fit dialog under Input tab, for Y Error choose this weighting column, and under Fit Control tab for Errors as Weight choose Direct Weighting.
Meili
OriginLab Tech Support |
|
|
|
Keq
Germany
4 Posts |
Posted - 05/20/2016 : 12:58:24 PM
|
thank you,
i think i found a good solution for my problem. just another question: is it possible in the same case to add a condition bending of the polynom? like a condition for the second derivative, so that the polynom has to be bended to the left? and if yes how has that condition to be formulated. i cant find anything in the examples of the conditions |
|
|
|
meili_yang
103 Posts |
Posted - 05/27/2016 : 09:05:30 AM
|
Hi Keq,
I am not sure I totally understand what you mean. Could you possibly paste your data and fitting function, along with description how you want to fit in the post?
Thanks,
Meili
OriginLab Tech Support |
|
|
| |
Topic |
|
Polynomial Regression with " must conatin values"