CODEWITHSUNDEEP
pythonpython-3.xnumpysorting
Krishnang K Dalal
Krishnang K Dalal

Reputation: 2556

Sorting Nd Numpy Array By Rows In Descending Order

I want to sort the following matrix by row values:

a = array([[1, 4, 6],
           [5, 3, 7],
           [8, 4, 1]])

as

a = array([[6, 4, 1],
           [7, 5, 3],
           [8, 4, 1]])

I'm able to get the sorting indices using np.argsort(-a) which returns the following matrix of indices:

>>> a_idx = np.argsort(-a)
array([[2, 1, 0],
       [2, 0, 1],
       [0, 1, 2]])

but using these indices to rearrange the original matrix is not happening for me.

>>> a[a_idx]
array([[[8, 4, 1],
        [5, 3, 7],
        [1, 4, 6]],

       [[8, 4, 1],
        [1, 4, 6],
        [5, 3, 7]],

       [[1, 4, 6],
        [5, 3, 7],
        [8, 4, 1]]])

How to efficiently accomplish such a task? Thanks a lot in advance.

Upvotes: 2

Views: 908

Answers (2)

shaik moeed
shaik moeed

Reputation: 5785

Try this, using .take_along_axis() method,

>>> a = np.array([[1, 4, 6],
                  [5, 3, 7],
                  [8, 4, 1]])   
>>> a_idx = np.argsort(-a)

Output:

>>> np.take_along_axis(a, a_idx, axis=1)        
array([[6, 4, 1],
       [7, 5, 3],
       [8, 4, 1]])

Upvotes: 4

Abercrombie
Abercrombie

Reputation: 1086

Another method

import numpy as np
a = np.array([[1, 4, 6],
           [5, 3, 7],
           [8, 4, 1]])
print(np.sort(-a,axis=1)*-1)

Upvotes: 2

Related Questions