Sort numpy matrix row values in ascending order

Question

I have this following numpy matrix that I want to sort in ascending order based on the 3rd column values.

[[  3.05706500e+06   4.98000000e+01  -2.62500070e+01  -9.38135544e+01]
 [  3.05706600e+06   4.98000000e+01  -3.00000056e+01  -9.38135544e+01]
 [  3.05706700e+06   4.98000000e+01  -3.37500042e+01  -9.38135544e+01]
 [  3.05706800e+06   4.98000000e+01  -3.75000028e+01  -9.38135544e+01]]

This is the matrix I actually want.

[[  3.05706800e+06   4.98000000e+01  -3.75000028e+01  -9.38135544e+01]
 [  3.05706700e+06   4.98000000e+01  -3.37500042e+01  -9.38135544e+01]
 [  3.05706600e+06   4.98000000e+01  -3.00000056e+01  -9.38135544e+01]
 [  3.05706500e+06   4.98000000e+01  -2.62500070e+01  -9.38135544e+01]]

How do I do this using just numpy? Any help would be appreciated. Thanks!

possible duplicate of [sorting arrays in numpy by column](http://stackoverflow.com/questions/2828059/sorting-arrays-in-numpy-by-column) — svural, Sep 19 '14 at 21:57

Abhijit · Accepted Answer · 2012-04-09T20:29:42.170

Given your array

>>> arr
array([[  3.05706500e+06,   4.98000000e+01,  -2.62500070e+01,
         -9.38135544e+01],
       [  3.05706600e+06,   4.98000000e+01,  -3.00000056e+01,
         -9.38135544e+01],
       [  3.05706700e+06,   4.98000000e+01,  -3.37500042e+01,
         -9.38135544e+01],
       [  3.05706800e+06,   4.98000000e+01,  -3.75000028e+01,
         -9.38135544e+01]])

you can simply use numpy.sort with axis=0 to sort it as desired

>>> numpy.sort(arr,axis=0)
array([[  3.05706500e+06,   4.98000000e+01,  -3.75000028e+01,
         -9.38135544e+01],
       [  3.05706600e+06,   4.98000000e+01,  -3.37500042e+01,
         -9.38135544e+01],
       [  3.05706700e+06,   4.98000000e+01,  -3.00000056e+01,
         -9.38135544e+01],
       [  3.05706800e+06,   4.98000000e+01,  -2.62500070e+01,
         -9.38135544e+01]])
>>>

I believe my previous answer was wrong as I misunderstood the question. Here is the correct answer

>>> arr[arr[:,2].argsort()]
array([[  3.05706800e+06,   4.98000000e+01,  -3.75000028e+01,
         -9.38135544e+01],
       [  3.05706700e+06,   4.98000000e+01,  -3.37500042e+01,
         -9.38135544e+01],
       [  3.05706600e+06,   4.98000000e+01,  -3.00000056e+01,
         -9.38135544e+01],
       [  3.05706500e+06,   4.98000000e+01,  -2.62500070e+01,
         -9.38135544e+01]])
>>>

You are sorting every column, not with respect to 3rd column. The output is right, but I am not that this is the method that OP was looking for. — Akavall, Apr 09 '12 at 20:00
@Akavall Thanks, I was yet to try out the solution. Thanks to you both! — prrao, Apr 10 '12 at 14:58
I had to also specify the axis, i.e. sample[:,2].argsort(axis=0) — GregNash, Jun 21 '13 at 21:55
Hello. Could you please say how can I sort every matrix column in descending order? — Abrcd18, Nov 16 '20 at 11:45

Sort numpy matrix row values in ascending order

1 Answers1