Numpy reshaping 2D to 3D: moving columns to "depth"

Question

Question

How do I reshape a 2D matrix to a 3D matrix where columns are moved "depth-wise"?

I want array a to look like array b

import numpy as np

a = np.array([
                [1,1,1,1,1,1,2,2,2,2,2,2,3,3,3,3,3,3], 
                [1,1,1,1,1,1,2,2,2,2,2,2,3,3,3,3,3,3], 
                [1,1,1,1,1,1,2,2,2,2,2,2,3,3,3,3,3,3], 
                [1,1,1,1,1,1,2,2,2,2,2,2,3,3,3,3,3,3]
            ])

b = np.array([
                [[1,1,1,1,1,1], 
                 [1,1,1,1,1,1], 
                 [1,1,1,1,1,1], 
                 [1,1,1,1,1,1]],
                [[2,2,2,2,2,2], 
                 [2,2,2,2,2,2], 
                 [2,2,2,2,2,2], 
                 [2,2,2,2,2,2]],
                [[3,3,3,3,3,3], 
                 [3,3,3,3,3,3], 
                 [3,3,3,3,3,3], 
                 [3,3,3,3,3,3]],
            ])

What I tried to do

The result is not the result that I want to achieve

x = np.reshape(a, (a.shape[0], -1, 3))

I've also tried the following: 1) hsplit a into 3 sets 2) tried to dstack these sets but doesn't produce wanted result

b = np.hsplit(a,3)
c = np.dstack([b[0],b[1],b[2]])

Answer 1

this should do:

import numpy as np

a = np.array([
                [1,1,1,1,1,1,2,2,2,2,2,2,3,3,3,3,3,3], 
                [1,1,1,1,1,1,2,2,2,2,2,2,3,3,3,3,3,3], 
                [1,1,1,1,1,1,2,2,2,2,2,2,3,3,3,3,3,3], 
                [1,1,1,1,1,1,2,2,2,2,2,2,3,3,3,3,3,3]
            ])

b = np.array([
                [[1,1,1,1,1,1],
                 [1,1,1,1,1,1],
                 [1,1,1,1,1,1],
                 [1,1,1,1,1,1]],
                [[2,2,2,2,2,2],
                 [2,2,2,2,2,2],
                 [2,2,2,2,2,2],
                 [2,2,2,2,2,2]],
                [[3,3,3,3,3,3],
                 [3,3,3,3,3,3],
                 [3,3,3,3,3,3],
                 [3,3,3,3,3,3]],
            ])

a_new = np.swapaxes(a.reshape(a.shape[0], 3, -1), 0, 1)

np.array_equal(a_new, b)

-> True

Answer 2

You just need transpose which is T and reshape

a.T.reshape(3,4,6)

or

a.T.reshape(b.shape)

Out[246]:
array([[[1, 1, 1, 1, 1, 1],
        [1, 1, 1, 1, 1, 1],
        [1, 1, 1, 1, 1, 1],
        [1, 1, 1, 1, 1, 1]],

       [[2, 2, 2, 2, 2, 2],
        [2, 2, 2, 2, 2, 2],
        [2, 2, 2, 2, 2, 2],
        [2, 2, 2, 2, 2, 2]],

       [[3, 3, 3, 3, 3, 3],
        [3, 3, 3, 3, 3, 3],
        [3, 3, 3, 3, 3, 3],
        [3, 3, 3, 3, 3, 3]]])