Sorting values in an array: 'reverse' is an invalid keyword argument for this function

Question

I try to sort values in an array in descending order. If I try it in ascending order it works, but when i do it descending I get an error.

e = np.array([[5.,3.,8.],[6.,7.,1.],[4.,8.,2.]])
e.sort()

result:

e = array([[3.,5.,8.],[1.,6.,7.],[2.,4.,8.]])

now in reverse order:

  e.sort(reverse=True)

result:

TypeError: 'reverse' is an invalid keyword argument for this function

I also tried e.sort(key=itemgetter(1)) after from operator import itemgetter but the same error appears ('reverse' is replaced by 'key').

Why is this the case? Why does it not work? Why this error (this is the way to use key or reverse right)?

I've found my result here: link f = -np.sort(-e) worked perfect! — S Verhoef
– S Verhoef, Commented Aug 17, 2017 at 15:27

f1u77y · Accepted Answer · 2017-08-17 14:27:45Z

2

You couldn't use the key or reverse keyword arguments according to numpy documentation. You can sort the array in ascending order and then reverse it with the [::-1] slice or use the reversed() view.

answered Aug 17, 2017 at 14:27

f1u77y

1614 bronze badges

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

Comments

Scott Newson · Accepted Answer · 2017-08-17 14:27:08Z

0

Try using fliplr after sitting in ascending order https://docs.scipy.org/doc/numpy-dev/reference/generated/numpy.fliplr.html#numpy.fliplr

answered Aug 17, 2017 at 14:27

Scott Newson

2,8352 gold badges25 silver badges26 bronze badges

1 Comment

S Verhoef Over a year ago

I've also some nan values in my dataset which should stay at the end of the array. By using fliplr these would be put in front. Otherwise this would be perfect.

mgilson · Accepted Answer · 2017-08-17 14:27:16Z

0

Why is this the case? Why does it not work?

Simply because numpy's sort function doesn't supply that functionality.

One way to "fix" this problem is to argsort on the column of interest. You can then reverse the output and use that to index e to get the items sorted as you wish:

>>> import numpy as np
>>> e = np.array([[5.,3.,8.],[6.,7.,1.],[4.,8.,2.]])
>>> e
array([[ 5.,  3.,  8.],
       [ 6.,  7.,  1.],
       [ 4.,  8.,  2.]])
>>> col = 0
>>> idx_asc = np.argsort(e[:, col])
>>> idx_desc = idx_asc[::-1]
>>> e[idx_desc]
array([[ 6.,  7.,  1.],
       [ 5.,  3.,  8.],
       [ 4.,  8.,  2.]])

Note that this approach is less flexible (and less convenient) than being able to pass a key and reverse=True. Specifically, if you want to sort on multiple columns to break ties, this approach fails.

answered Aug 17, 2017 at 14:27

mgilson

312k70 gold badges656 silver badges722 bronze badges

2 Comments

S Verhoef Over a year ago

No column is of special interest, all rows should be sorted indivudually.

mgilson Over a year ago

@SVerhoef -- I guess I don't understand what you're getting at with they key=itemgetter(1) then ...

Collectives™ on Stack Overflow

Sorting values in an array: 'reverse' is an invalid keyword argument for this function

3 Answers 3

Comments

1 Comment

2 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

1 Comment

2 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related