| Author |
 Topic  |
|
|
zazza123
2 Posts |
 Posted - 04/16/2018 : 08:11:05 AM
|
Origin Ver. and Service Release: OriginPro 2018
Operating System: Win10 pro
Good morning, I'm writing here because I'm a new Origin user and I have little confidence with the software. I have an issue which I'm not able to solve, regarding a non-linear fit procedure.
Before, I have to sum up what I'm doing and what I did till now with Excel. Basically I have a calibrated sample (NIST traceable) which let's say give me the value of a physical quantity M in the range from 0 to 4 (arbitrary units). For my purposes and daily applications, I can measure the quantity M even to values of 7, because my instruments allow me to do so. Now, in order to calibrate my instruments to such values I perform this task: firstly I only measure the calibrated sample and secondly one I add to the calibrated sample another sample that I can make in the lab which is 2.6 in units. In this way I can extend the range of my "calibration measurement" instead of 0 to 4, from 0 to 6.5. (The values are fictional).
Set1(ref)
0
1
2
3
4
2.6 (0 + 2.6)
3.6 (1 + 2.6)
4.6 (2 + 2.6)
5.6 (3 + 2.6)
6.6 (4 + 2.6)
Set2(measured value with my instrument)
0.02
1.06
2.18
3.08
4.06
2.48
3.69
4.46
5.57
6.75
The second step is to measure and store the values that I obtain with my instrument for such "nominal values" together with their uncertainties. To fit the data I have a non-linear function, which is a linear function which has a saturation at high levels (there's a bending or knee of the curve). My function has in total 2 parameters.
In my model function I put the values of Set1 and I obtain the "theoretical values". Then I perform the Chi-square test (squared difference between measured value vs therotetical value) and finally I use the solver function of excel asking to minimize the sum of the chi-squared, changing the 2 parameters plus the value that I measured for my internal sample, namely 2.6
In this way, the solver change these 3 values and find the parameters and also the "real" value of my internal sample, which suppose after the fit is 2.32, thus changing the second part of set1 in this way:
Set1(ref+internal sample)
0 + 2.32 = 2.32
1 + 2.32 = 3.32
2 + 2.32 = 4.32
3 + 2.32 = 5.32
4 + 2.32 = 6.32
Unfortunately Excel is not able to retrieve me the standard error of the parameters, which is something that I need, of course.
I hope you can give me some hints or advices to solve this problem... I know that my independent variables are actually not really independent, since just set1 is truly independent, while set2 depends in a way from this internal sample value, which is not included in the equation for the fit.
At this moment I don't know how to parametrize my data in order to obtain the desired outcome. Hope you could give me a hint.
Best regards |
|
|
yuki_wu
896 Posts |
Posted - 04/17/2018 : 04:42:04 AM
|
Hi,
I suppose you already know this is Origin Forum, right? Do you want to get the value of standard error after performing non-linear fitting in Origin?
Regards,
Yuki
OriginLab
|
 |
|
|
zazza123
2 Posts |
Posted - 04/17/2018 : 10:07:14 AM
|
@yuki_wu I'm sorry but I did not get your question. I'd like to get the best estimate of my fit parameters, and of course I know that Origin can retrieve the SE.
The problem is that I don't know how to set origin, to perform a task which is very easy to do in Excel.
|
|
|
|
yuki_wu
896 Posts |
|
| |
Topic |
|
|
|
Suggestions for fit - solver Excel