| Author |
 Topic  |
|
|
CRigby
South Africa
1 Posts |
 Posted - 06/18/2017 : 05:48:45 AM
|
Origin Ver. and Service Release: OriginPro 2015 Sr2 b9.2.272
Operating System: Windows 10
Good day.
I have a data set (given below) which is to be used to calibrate the readout of a wavelength tunable laser, which should be pretty linear. When applying a linear fit, the residuals appear to form a gaussian or bell shape.
I would like to be able to account for this in my fit and thereby reduce the standard deviation as I believe this may be a real artefact of my system.
In short I would like to know how one could fit a gaussian with a slanted base. I have tried the custom equation fitting by simply adding the equations for a line and an amplitude gaussian together, but I cannot get a fit with reasonable parameters.
I have tried for more than a week now with various combinations, but now need to get this done soon. Either I have been doing something inherently wrong in the dialogs of the nonlinear fit option or I am on the wrong path entirely. If a kind soul could please help me with the correct equation and parameter values I would be greatful.
X, Y,
22690.1928267043, 69086.24,
22684.785670402, 69080.8,
22684.0740016195, 69080.1,
22680.8964796708, 69076.91,
22678.6666749217, 69074.71,
22676.2146531386, 69072.25,
22672.4923786984, 69068.53,
22671.7455004067, 69067.8,
22670.757619993, 69066.81,
22665.5056185862, 69061.58,
22656.7802748621, 69052.92,
22648.1401256736, 69044.32,
22641.1386319183, 69037.29,
22639.9114833797, 69036.07,
22638.7838439886, 69034.92,
22631.8639897396, 69027.97,
22629.9172801845, 69026.02,
22628.6746546371, 69024.77,
22621.7996148322, 69017.87,
22619.1168497017, 69015.19,
22617.7114023542, 69013.78,
22610.9292187472, 69006.99,
22595.1570466788, 68991.2,
22586.8336476946, 68982.9,
22691.750493756, 69087.79,
22687.888409276, 69083.91,
22568.1702048066, 68964.12,
22559.811248376, 68955.7,
22573.6971957486, 68969.67,
22581.9899552147, 68978.04,
|
|
|
Shirley_GZ
China
Posts |
Posted - 06/19/2017 : 11:29:18 PM
|
Hi CRigdy,
I am afraid that you have misunderstood the definition of residual plots.
Actually, residual plot is not plotted from residual of Y vs. X, you can see the detailed explanations on this page:
http://www.originlab.com/doc/Origin-Help/Residual-Plot-Analysis
I have performed linear fit on your XY data, in the Residual Plot branch, I can only see the PDF(Probability Distribution Function) of the residual of Y is Gaussian. So, the combination you have tried is not applicable for your case, I think.
Please let me know if you have any other doubt.
Thanks,
Shirley
Originlab Technical Service Team |
 |
|
| |
Topic |
|
|
|
Help with linear and gaussian fit combination