uwot::umap_transform not producing same results as those in python implementation about uwot

[moldach]

Wow that was a quick and very accurate response. Converting to a factor solved the issue - thanks!

[jlmelville]

I'd say that the py$y array, because it contains numeric 1s and 0s, is being interpreted as containing numeric values, not categorical values. Convert them to a factor and you should see the results you are expecting:

sup_embed_R <- umap(py$X, y = factor(py$y))