| Author |
 Topic  |
|
|
tsungtofu
Taiwan
1 Posts |
 Posted - 04/15/2025 : 03:48:18 AM
|
Hello,
I'm performing data using global fitting to simultaneously fit both xx/xx(H=0) and xy/xx(H=0) components. I've obtained different fitting results that I'm unsure how to evaluate. Below are two different fitting sets:
My questions are:
1. How can I determine which of these two global fitting results is more reliable?
2. When multiple parameters show "Dependency" values close to 1, how should I evaluate the fitting results?
3. What is the significance of missing standard errors (shown as "--") for some parameters, and how does this affect the reliability of the results?
Thanks for you help!
-----------
Origin Ver. : Origin 2024b (64-bit) 10.1.5.132
Operating System: Windows 11 Enterprise Build |
|
|
YimingChen
1647 Posts |
Posted - 04/15/2025 : 09:20:28 AM
|
Hi,
Here are the answers to your questions:
1. How can I determine which of these two global fitting results is more reliable?
It’s difficult to determine which fitting is more reliable, as convergence may occur by settling into a local minimum in the parameter space. You may want to try different initial values for the fitting parameters. Please refer to the page below for more information.
https://www.originlab.com/doc/Origin-Help/The_Reason_Why_Fail_to_Converge#Relative_vs_absolute_minima
2. When multiple parameters show "Dependency" values close to 1, how should I evaluate the fitting results?
It’s likely that your fitting function is overparameterized, which results in a dependency value close to 1. You may need to fix some of the parameters. Please refer to the page below.
https://www.originlab.com/doc/Origin-Help/Model_Diagnosis_Using_Dependency_Values
3. What is the significance of missing standard errors (shown as "--") for some parameters, and how does this affect the reliability of the results?
This could be due to the fitted values of parameters N1, N2, N3 being too large, causing SE to exceed the Origin's maximum allowable value. Please refer to this page
https://www.originlab.com/doc/Quick-Help/goodfit-for-large-small-param
Thank you
James |
 |
|
| |
Topic |
|
|
|
Evaluating the Reliability of Global Fit Result