Leland McInnes

It may just be a memory issue -- as in not enough of it. UMAP can be pretty memory hungry when doing nearest neighbor computations, and depending on that dataset...

Worst case, try downgrading to an earlier pynndescent version (0.5.1 should work). I'm still trying to figure out the right way to deal with all of this.

It wasn't an explicit design choice to avoid it. I would have hoped that the ``plt.colorbar`` would have worked. My matplotlib is only good enough to do some bits however....

I would get @timsainb to weigh on more detailed aspects. PramatericUMAP doesn't support combining models in the same way, so you will need to simply come up with a reasonable...

If you want to isolate a cluster you'll need to run a clustering algorithm over the embedding and use the results of that. Personally I recommend HDBSCAN, but there are...

If you don't specify the metric as precomputed UMAP will interpret the input as data and use the default metric (euclidean) to compute distances between rows. So the result would...

There are reasons that this could happen, and measuring euclidean distance between rows of a distance matrix is not inherently a bad thing to be doing -- you would just...

Sorry for the (very) slow reply. I have not actually seen other applications doing this. That doesn't mean they don't exist, just that I have no come across them. As...

It *ought* to. The package ``__init__`` from the [following line onward](https://github.com/lmcinnes/umap/blob/ec9832cbdb7f106a727e6bbd4c54c78fae30f26f/umap/__init__.py#L36) shoudl ensure that, one way or another, some ``__version__`` attribute gets set. It is not obvious to me why...

It gets used if you ever apply a transform to new data where we need to query the nearest neighbors of the new data in the training set (and the...