Join GitHub today
GitHub is home to over 36 million developers working together to host and review code, manage projects, and build software together.Sign up
REF DRYing fit methods #1615
It looks like we are adding or copying many fit functions just because they need to choose the results class. In some cases they also have special code to calculate start_params.
similar use case:
(while looking at the missing NegativeBinomial.fit_regularized #1454)
I'm running into this again for adding another fit_method (
The only difference is in the Results and ResultsWrapper classes
Yes, this was part of the refactor that moved these into types of discrete choice models since the return classes have a lot of model-specific code.
I'm +1 on a