MNIST: ValueError

[kerighan]

There's an issue when running umato on MNIST dataset.

import umato 
from sklearn.datasets import fetch_openml

mnist = fetch_openml("mnist_784", version=1)
X = mnist.data

emb = umato.UMATO(hub_num=20).fit_transform(X)

results in

ValueError: total data # (70000) != total embedded # (69992)!

[hj-n]

Thank you for the bug report.
I succeed to reproduce the error; it will be handled soon

[cakiki]

This is just to confirm that I'm running into this same error on my own dataset.

ValueError: total data # (312860) != total embedded # (312781)!

[LeandroLuque]

Hi. After to install manually the package (it is not longer available on pip) and try to run the example. This example doesn't work.
When I try UMATO with my data, I received this msg:

ValueError: total data # (26170) != total embedded # (26109)!

I hope you can fix this issue to test your method.

[cakiki]

@LeandroLuque It is still available: https://pypi.org/project/umato/

[da-bar]

Any updates on this? I am running on my data and getting the same error.

[hj-n]

Sorry for the late reply.
The issue has been resolved.
Refer to #16 written by @SyphonArch.