First NMF function

[ismav]

This should hopefully be the right pull request!

[yarikoptic]

Great. But since we are doing it within scikit-learn, better to make this nmf as a module under sklearn, right? so move it there, add also an __init__.py file so Python recognizes it as a submodule

[yarikoptic]

actually better as sklearn/decomposition/sparse_nmf.py I guess

[ismav]

Well, we will have implementations which work in Spark clusters as well as
GPU machines.
So, not sure if we want it in the sklearn branch.
For example:

https://github.com/arnlaugsson/sparkIt

Okay, will do the tests similar to the NMF function in scikits-learn.

On Sun, May 3, 2015 at 10:27 AM, Yaroslav Halchenko <
notifications@github.com> wrote:

In nmf/sparse_nmf.py
#2 (comment):

@@ -0,0 +1,172 @@
+""" Sparse Nonnegative matrix factorization

actually better as sklearn/decomposition/nmf

—
Reply to this email directly or view it on GitHub
https://github.com/scicubator/scikit-learn/pull/2/files#r29556542.