extend cluster function api

[andrewheusser]

Add the following to hyp.tools.cluster:

• Align flag which aligns data before running clustering alg
• model flag to specify dimensionality reduction alg used if ndims is not None
• model_params flag to specify model parameters for dimensionality reduction model

@rarredon any interest? similar to what you helped with during the moz sprint :)

[jeremymanning]

Also: we should add the normalize flag

[rarredon]

Sounds good. I should be able to get those implemented this week, maybe even today.

[jeremymanning]

awesome-- thanks @rarredon!

[rarredon]

OK, I implemented the new kwargs but I'm getting an error when align=True. It seems kmeans.fit(x) expects x to be a 1 or 2 dimensional array for example, something like,

array([[1, 2, 3],
       [4, 5, 6],
       [7, 8, 9]])

but tools.align returns a 3 dimensional array; something like

[array([[1, 2, 3],
        [4, 5, 6],
        [7, 8, 9]])]

Do you guys have any ideas for how to handle this?

[andrewheusser]

I think you want something like this before passing the data to kmeans.fit:

    if type(x) is list:
        x = np.vstack(x)

This should work, let me know!