How to slice multidimensional array with Numpy, multiple columns?

Question

I am generating multidimensional array of different sizes, though they'll all have an even number of columns.

>> import numpy as np
>> x = np.arange(24).reshape((3,8))

Which results in:

array([[ 0,  1,  2,  3,  4,  5,  6,  7],
       [ 8,  9, 10, 11, 12, 13, 14, 15],
       [16, 17, 18, 19, 20, 21, 22, 23]])

I am able to slice with numpy and get the first column in an array:

>> newarr = x[0:,0:2]

array([[ 0,  1],
       [ 8,  9],
       [16, 17]])

However, I want to have one array that is just a list of the columns where column 1 and 2 are together, 3 and 4 are together, and so on.. For example:

array([[[ 0,  1],
       [ 8,  9],
       [16, 17]],
       [[ 2,  3],
       [10, 11],
       [18, 19]],
       etc....]
)

This code below works but it's clunky and my arrays are not all the same. Some arrays have 16 columns, some have 34, some have 50, etc.

>> newarr = [x[0:,0:2]]+[x[0:,2:4]]+[x[0:,4:6]]

[array([[ 0,  1],
       [ 8,  9],
       [16, 17]]), array([[ 2,  3],
       [10, 11],
       [18, 19]])]

There's got to be a better way to do this than

newarr = [x[0:,0:2]]+[x[0:,2:4]]+[x[0:,4:6]]+...+[x[0:,n:n+2]]

Help!

With numpy slicing there is an option called the step which is formatted as such [start:stop:step]. One example of this is that if you wanted to get every nth column you could do [::n]. This seems relevant to what you are trying to do. — Duncan McFarlane
– Duncan McFarlane, Commented Jul 9, 2019 at 21:17

Hu Xixi · Accepted Answer · 2019-07-10 02:37:56Z

1

My idea is adding a for loop:

slice_len = 2
x_list = [x[0:, slice_len*i:slice_len*(i+1)] for i in range(x.shape[1] // slice_len)]

Output:

[array([[ 0,  1],
        [ 8,  9],
        [16, 17]]), array([[ 2,  3],
        [10, 11],
        [18, 19]]), array([[ 4,  5],
        [12, 13],
        [20, 21]]), array([[ 6,  7],
        [14, 15],
        [22, 23]])]

answered Jul 10, 2019 at 2:37

Hu Xixi

2,1972 gold badges26 silver badges32 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

Tuffknitz Over a year ago

Perfect! This is exactly what I was trying to do.

Hu Xixi Over a year ago

@Tuffknitz That's good to be helpful. But you should be careful that when the column size is odd, the number in range should plus one, i.e. range(x.shape[1] // slice_len + 1) so maybe there also need a if ... else....

Tuffknitz Over a year ago

Thank you for the comments :) I have a check early on in the code that validates this array - if it's an odd number of columns then something is wrong and it will throw an exception.

Collectives™ on Stack Overflow

How to slice multidimensional array with Numpy, multiple columns?

1 Answer 1

3 Comments

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Related