GLM.fit maxiter=0 incorrect #1728

Open
josef-pkt opened this Issue Jun 2, 2014 · 1 comment

Projects

None yet

1 participant

@josef-pkt
Member

GLM fit still does a lot of initialization calculations that are not necessary if maxiter=0.

I'm using maxiter=0 with start_params=params to get a full results instance.

I think the error is in missing offset and exposure in the initialization
mu = self.family.fitted(np.dot(wlsexog, start_params)) see #1486

This was referenced Jun 4, 2014
@josef-pkt josef-pkt closed this in #1821 Jul 9, 2014
@josef-pkt
Member

maxiter=0 and method='irls' returns results with params is None

see #2393 for current case, working towards a generic solution

@josef-pkt josef-pkt reopened this May 13, 2015
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment