| Author |
 Topic  |
|
|
helporigin8
29 Posts |
 Posted - 03/07/2017 : 07:50:47 AM
|
Hi,
I would like to know is it possible to somehow generate error bars for function that was used for fitting experimental data regarding to the data errors and the goodness of fit?
Thank you,
A. |
|
|
arstern
USA
237 Posts |
Posted - 03/07/2017 : 11:32:01 AM
|
Hi Aleksandra,
To show the error of your fitted curve you can plot a confidence band. You can also show a prediction band which shows the estimated standard error of prediction. This tutorial shows you how to set the graph to plot the confidence and prediction band https://www.youtube.com/watch?v=WDAThxAXRDk. You can also read more about confidence and prediction intervals here: http://www.originlab.com/doc/Origin-Help/Fitted_Curve_Plot_Analysis . A residual plot may be useful as well to estimate the error. The residual plot and residual data should automatically be provided in the worksheet when you create a fitted curve.
Hope this helps. If I did not address your question correctly, please let us know.
--Aviel Stern
OriginLab |
Edited by - arstern on 03/07/2017 11:34:08 AM |
 |
|
|
helporigin8
29 Posts |
Posted - 03/09/2017 : 05:31:37 AM
|
Thank you very much. I fitted my data and program generated the prediction and confidence bands. But I am worried about confidence bands that are equal to the cumulative fit as would be case if my data error bars are =0. Do you know reason for this? And second, I would like to estimate the prediction and confidence bands for the peak 2 alone, without contributions of the peaks 1 and 3, if it is possible to do automatically using Origin, same as for cumulative fit?
|
|
|
|
arstern
USA
237 Posts |
Posted - 03/09/2017 : 2:32:49 PM
|
Hi,
From the graph that was posted, I was able to extract the data points and create a fitted curve with a confidence and prediction band. Below is the graph that was produced. I am not sure why your confidence band equals the cumulative fit curve. If you could send us your .opj file to technical support - tech@originlab.com, then I can evaluate your data and your graph further.
To address determining the confidence band/cumulative fit from each peak, you will have to select only one peak rather than all peaks when you create the fit curve from Analysis --> Peaks and Baseline --> Multiple Peak fit.
-Aviel Stern
OriginLab |
|
|
| |
Topic |
|
|
|
Error bars for fitting function