Leland McInnes

Numba is the the main dependency that tends to be a blocker. I am not aware that they support 3.13 yet, but likely will soon.

That looks like the sort of results you can get when an index column made it into the data by accident (I have certainly done this myself once or twice)....

I think the real catch here is the DTW is a pretty expensive metric to compute, so you are going to face a bit of an uphill battle no matter...

``compressed=True`` indexes are, unfortunately, not updateable; a bunch of the information needed for the update is dropped to compress the whole thing down to be smaller.

The other linkage implementations use ``scipy.cluster.hierarchy.linkage`` which calls ``pdist`` which has default values (in the validator) for the extra args required for seuclidean and mahalanobis. For consistency the right thing...

I'll have to figure out what has gone astray here. It isn't that obvious immediately. Thanks for at least documenting the workaround.

Yes -- this is definitely a quirk of the default scikit-learn API. I actually like your workaround here. It means that ParametricUMAP will diverge in behaviour from standard UMAP, but...

This is potentially useful, although it does result in longer runtimes for umap as the lazy loading occurs at runtime. This may be a little unexpected for some users. Let...

Sample weighting is something I have given quite a bit of thought to over the years, but I didn't manage to come up with a truly satisfactory solution, so it...

I don't have any immediate solutions to this. Increasing distances in the distance matrix only gets around direct connection, but potentially if there is a path that connects the two...