Closed micahfolsom closed 2 years ago
Can you print df.index
in param_dataframe()
?
Looks like it's due to more than one underscore in some of the parameters, so tuple(p.split("_")) for p in df.index
has different numbers of elements.
Maybe we need to ensure that we enforce non-empty prefixes ending in an underscore when models are instantiated, since the sort_by_model
statement here requires this convention. (DeprecationWarning
for now?)
This will be fixed by https://github.com/lbl-anp/becquerel/pull/312 which splits the "model_paramname" string on only the first ``, instead of all of them, so it doesn't trip up on param names with underscores
@micahfolsom did #312 solve this? If so, please close this issue.
While adding an example to the
fitting
notebook for this PR, I ran into a weird error. As far as I can tell I'm doing the same exact thing as the double Gauss example earlier in the notebook. The code:For some reason, the fit works fine and yields the expected result. The failure occurs on
fitter.custom_plot()
:Any ideas what's going on here?