Open dhops opened 3 years ago
As far as I understand by
in pyGAM only implements factors. So if you have array X
and use s(3, by=0)
it internally fits to the datapoints X[3] * X[0]
, where X[0]
are the actual numbers as floats of your 0th feature column. So if you're planning on using X[0]
as a categorical variable list (as in, fit a SEPARATE spline on feature X[3]
for every category provided by the categorical label in X[0]
) I don't think this is supported in pyGAM. This would be a feature I would appreciate as well @dswah :)
Hi, I'm just checking on the status of "by-variables" in pyGAM. I saw that "by=" is an option, but when I create a term with by, for example s(3, by=0), the term just becomes s(3).