Flatten np.ndarray in customized order

Question

I have a np.ndarray with the shape (24, 3). I want to flatten this array but in a rathor unusual way. I would like to have [0:8, 0] then [0:8, 1] then [0:8, 2] then [8:16, 0] and so on.

Of course i could do it the brute force way but maybe there is a more elegant and efficient solution to this problem.

new_array = np.array([])
np.append(new_array, old_array[0:8, 0])
np.append(new_array, old_array[0:8, 1])
np.append(new_array, old_array[0:8, 2])

np.append(new_array, old_array[8:16, 0])
np.append(new_array, old_array[8:16, 1])
np.append(new_array, old_array[8:16, 2])

np.append(new_array, old_array[16:24, 0])
np.append(new_array, old_array[16:24, 1])
np.append(new_array, old_array[16:24, 2])

Does this do what you want `np.array(np.array_split(a, 3)).swapaxes(1,2).ravel()`? — yatu, Jun 01 '19 at 19:09

score 1 · Accepted Answer · answered Jun 01 '19 at 19:13

1

Reshape, permute and reshape -

n = 8 # cut length along first axis
new_array = old_array.reshape(-1,n,old_array.shape[1]).swapaxes(1,2).ravel()

answered Jun 01 '19 at 19:13

Divakar

218,885
19
262
358

Flatten np.ndarray in customized order

1 Answers1