LinearFit uncertainty

[Kale Letterman]

Hi!

I wanted more so to confirm that the following doesn't work. I have x,y data and uncertainty of the y values too in a Linear fit. I wanted to ask if I can somehow take these uncertainties into consideration in the Lmfit package.

Thanks in advance

Best Kale

[Matt Newville]

You can take uncertainties in y data into account when doing a fit with lmfit.  How you do that depends on what part of the library you are using. For example, if you are fitting data with the Model class, then use 'weight=1./y_uncertainty'.

If you need any more specific help, ask questions about code that you post.

--
You received this message because you are subscribed to the Google Groups "lmfit-py" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lmfit-py+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lmfit-py/db08eea2-a3ff-4d7c-a484-6c30fc168f44n%40googlegroups.com.

[Laurence Lurio]

You use the weights keyword argument.  The weights should be 1/dy with dy the uncertainty in y.

e.g.

import numpy as np
from lmfit.models import LinearModel
x = np.array([1,2,3.0])
y = np.array([4,9,16.1])
dy = np.array([1,4,.2])
result = LinearModel().fit(y,x=x,weights=1/dy)
print(result.fit_report())

--

[Kale Letterman]

That's perfect, thanks!