How do I stack multiple columns from one single array on top of each other in Python?

Question

For example, the shape of the array is 5,4.

a = np.random.randint(10, size= (5, 4))

a = 
[[1 4 5 0]
 [3 1 5 1]
 [4 8 0 9]
 [8 1 5 8]
 [6 4 7 4]]

I want the array to be reshaped as:

a = 
[[1 4]
 [3 1]
 [4 8]
 [8 1]
 [6 4]
 [5 0]
 [5 1]
 [0 9]
 [5 8]
 [7 4]]

My original array size is around 200 GB and of shape 80000*480600. I have tried to use remap mode but it is very slow.

You need to add the code what you tried, so we know what is taking too much time. There is a duplicate post somewhere else. Please check this link and see if this helps. https://stackoverflow.com/questions/14476415/reshape-an-array-in-numpy — Vishwas, Jul 22 '19 at 13:39

Vicrobot · Accepted Answer · 2019-07-22T16:36:42.033

2

Use numpy.hsplit and numpy.concatenate:-

>>> a = np.random.randint(10, size= (5, 4))
>>> a
array([[8, 5, 8, 9],
       [9, 5, 6, 3],
       [5, 3, 8, 7],
       [9, 0, 9, 9],
       [0, 7, 8, 0]])
>>> t = np.hsplit(a, 2)
>>> t
[array([[8, 5],
       [9, 5],
       [5, 3],
       [9, 0],
       [0, 7]]), array([[8, 9],
       [6, 3],
       [8, 7],
       [9, 9],
       [8, 0]])]
>>> np.concatenate([t[0], t[1]])
array([[8, 5],
       [9, 5],
       [5, 3],
       [9, 0],
       [0, 7],
       [8, 9],
       [6, 3],
       [8, 7],
       [9, 9],
       [8, 0]])

edited Jul 22 '19 at 16:36

answered Jul 22 '19 at 13:53

Vicrobot

3,795
1
17
31

1

thank you for letting me know about np.hsplit! it was very helpful! – Sagnik Jul 23 '19 at 17:03

score 0 · Answer 2 · answered Jul 22 '19 at 18:03

0

hsplit and reshape also works

np.reshape(np.hsplit(a, 2), (-1,2))

Out[99]:
array([[1, 4],
       [3, 1],
       [4, 8],
       [8, 1],
       [6, 4],
       [5, 0],
       [5, 1],
       [0, 9],
       [5, 8],
       [7, 4]])

answered Jul 22 '19 at 18:03

Andy L.

24,909
4
17
29

score 0 · Answer 3 · answered Jul 22 '19 at 18:12

This can be done with a reshape and transpose (and a final reshape)

In [195]: arr = np.arange(20).reshape(5,4)                                                                   
In [196]: arr                                                                                                
Out[196]: 
array([[ 0,  1,  2,  3],
       [ 4,  5,  6,  7],
       [ 8,  9, 10, 11],
       [12, 13, 14, 15],
       [16, 17, 18, 19]])
In [197]: arr.reshape(5,2,2)                                                                                 
Out[197]: 
array([[[ 0,  1],
        [ 2,  3]],

       [[ 4,  5],
        [ 6,  7]],

       [[ 8,  9],
        [10, 11]],

       [[12, 13],
        [14, 15]],

       [[16, 17],
        [18, 19]]])
In [198]: arr.reshape(5,2,2).transpose(1,0,2)                                                                
Out[198]: 
array([[[ 0,  1],
        [ 4,  5],
        [ 8,  9],
        [12, 13],
        [16, 17]],

       [[ 2,  3],
        [ 6,  7],
        [10, 11],
        [14, 15],
        [18, 19]]])

Identifying the right transpose (or swapaxes) may require a bit of trial and error.

In [199]: arr.reshape(5,2,2).transpose(1,0,2).reshape(-1,2)                                                  
Out[199]: 
array([[ 0,  1],
       [ 4,  5],
       [ 8,  9],
       [12, 13],
       [16, 17],
       [ 2,  3],
       [ 6,  7],
       [10, 11],
       [14, 15],
       [18, 19]])

The equivalent with split and concatenate:

In [200]: np.concatenate(np.hsplit(arr,2), axis=0)                                                           
Out[200]: 
array([[ 0,  1],
       [ 4,  5],
       [ 8,  9],
       [12, 13],
       [16, 17],
       [ 2,  3],
       [ 6,  7],
       [10, 11],
       [14, 15],
       [18, 19]])

The transpose route should be faster.

Thanks a lot for the comment! – Sagnik Jul 23 '19 at 17:04 — Sagnik, Jul 23 '19 at 17:04

How do I stack multiple columns from one single array on top of each other in Python?

3 Answers3