Leland McInnes

I think this is possible, but tricky. The catch is that you want to to restrict the potential neighbors to be only those in the subset. This is hard to...

There are some newer papers, but I don't know of any widespread implementations. I do know of some ongoing work on adding a semi-supervised cluster extraction method to HDBSCAN that...

Ah, that's the multi-component spectral initialisation failing, because it doesn't support pre-computed metrics. I'm on vacation at the moment, but I can make a better error message when I get...

Hmm, I think the short answer is that the PCA diagnostic plot won't work for sparse data -- I don't think there are any tests for that particular combination. This...

To make that work you would need to write a ``gower_distance`` function as numba jit compile it. You can pass in a jit compiled function as a metric and it...

I think you can probably just add an interactive parameter, and have not too many other changes. It may be worth adding a few extra keyword parameters, specific to the...

Based on how Jaccard is defined I would code them as zeros. I presume, however, that you actually want to distinguish them from votes against -- which raises questions about...

UMAP is specifically designed around learning the structure of the data, and assuming it has some intrinsic structure that is ideally relatively low dimensional. Compressing the entire space is not...

This looks like impressive work, but for my own sake I would like to keep this version of umap as a python only library (for ease of maintenance and packaging)....

At best I think you could hope that the interpolated points are on a curve in the embedding space; that curve may be relatively straight, but it also may not....