CODEWITHSUNDEEP
pythonarraysnumpytensor
Alby
Alby

Reputation: 5742

How to reshape a multidimensional array from a particular arrangement to another arrangement?

I have a before_arr(2 x 3 x 4) multidimensional array. I want to turn it into a new_arr (3 x 2 x 4) with a specific arrangement pattern which I wrote below. 

import numpy as np

before_arr = np.array([
    [
     [0, 0, 0, 0],
     [0, 0, 0, 0],
     [0, 0, 0, 0],
     ],
    [
     [1, 1, 1, 1],
     [1, 1, 1, 1],
     [1, 1, 1, 1],
     ],
    ], dtype=float)

# what I want
new_arr = np.array([
    [
     [0, 0, 0, 0],
     [1, 1, 1, 1],
     ],
    [
     [0, 0, 0, 0],
     [1, 1, 1, 1],
     ],
    [
     [0, 0, 0, 0],
     [1, 1, 1, 1],
     ],
    ], dtype=float)

before_arr.reshape(3, 2, 4) doesn't give me what I want.

In [74]: before_arr.reshape(3, 2, 4)
Out[74]:
array([[[0., 0., 0., 0.],
        [0., 0., 0., 0.]],

       [[0., 0., 0., 0.],
        [1., 1., 1., 1.]],

       [[1., 1., 1., 1.],
        [1., 1., 1., 1.]]])

Upvotes: 2

Views: 44

Answers (2)

user3483203
user3483203

Reputation: 51165

Since you only need to swap the position of two axes, you can use np.swapaxes. This is probably the most straightforward approach to get your desired output.

before_arr.swapaxes(0, 1)


array([[[0., 0., 0., 0.],
        [1., 1., 1., 1.]],

       [[0., 0., 0., 0.],
        [1., 1., 1., 1.]],

       [[0., 0., 0., 0.],
        [1., 1., 1., 1.]]])

In a more general sense, you can use transpose to explicitly define an ordering of axes for your desired output, which would be beneficial if you needed to move more than one axis (although not strictly required here).

before_arr.transpose(1, 0, 2)

Upvotes: 1

Adam.Er8
Adam.Er8

Reputation: 13403

use zip to match respective rows.

try this:

import numpy as np

before_arr = np.array([
    [
     [0, 0, 0, 0],
     [0, 0, 0, 0],
     [0, 0, 0, 0],
     ],
    [
     [1, 1, 1, 1],
     [1, 1, 1, 1],
     [1, 1, 1, 1],
     ],
    ], dtype=float)

new_arr = np.array([*zip(*before_arr)])
print(new_arr)

Output:

[[[0. 0. 0. 0.]
  [1. 1. 1. 1.]]

 [[0. 0. 0. 0.]
  [1. 1. 1. 1.]]

 [[0. 0. 0. 0.]
  [1. 1. 1. 1.]]]

Upvotes: 2

Related Questions