Evgeni Burovski
Evgeni Burovski
Why it cannot be handled, Josef? And how does statsmodels select the knot vector?
> Each basis function adds a column to the design matrix. So we need approximately nobs basis functions (IIRC) if we have around nobs knots. The design matrix will then...
Thanks for the references! It's now clear that adding a `t=...` kwarg should be one of the first things to consider for a follow-up (where the current behavior would become...
Could you rebase on main or merge main into your branch? Some CI failures are probably fixed in main.
Mailing list discussion: https://mail.python.org/archives/list/[email protected]/thread/Z6ZN7VLD7GZV3XXVE3J52LF34MVKROBY/
I now realize that I never replied to @rgommers 's comment in that "let's deprecate interp2d" mailing list thread. Will reply here and link from the email thread then. The...
ISTM there are two separate topics here: one is specifically the GCV spline functionality and the other one is a bigger topic of an overall API consistency. The former can...
While I don't understand all details of what you're after, this sounds similar to gh-8781. For 1D `y`, you can do the masking manually, like (untested) ``` >>> mask =...
Ok, this example is I'd say well in the pandas-like land. (Note that you can do linear interpolation easily with interp1d). Maybe this being slow can be fixed in pandas...
A basic goal misalignment here is what is a nan exactly: is it 'not-a-number', per IEEE754 semantics, or is it 'missing' as is used in the pandas-land. While I appreciate...