I try to sort an array:
import numpy as np
arr = [5,3,7,2,6,34,46,344,545,32,5,22]
print "unsorted"
print arr
np.argsort(arr)
print "sorted"
print arr
But the output is:
unsorted
[5, 3, 7, 2, 6, 34, 46, 344, 545, 32, 5, 22]
sorted
[5, 3, 7, 2, 6, 34, 46, 344, 545, 32, 5, 22]
The array does not change at all
There are two issues here; one is that np.argsort returns an array of the indices which would sort the original array, the second is that it doesn't modify the original array, just gives you another. This interactive session should help explain:
In [59]: arr = [5,3,7,2,6,34,46,344,545,32,5,22]
In [60]: np.argsort(arr)
Out[60]: array([ 3, 1, 0, 10, 4, 2, 11, 9, 5, 6, 7, 8])
Above, the [3, 1, 0, ...] means that item 3 in your original list should come first (the 2), then item 2 should come (the 3), then the first (index is 0, item is 5) and so on. Note that arr is still unaffected:
In [61]: arr
Out[61]: [5, 3, 7, 2, 6, 34, 46, 344, 545, 32, 5, 22]
You might not need this array of indices, and would find it easier to just use np.sort:
In [62]: np.sort(arr)
Out[62]: array([ 2, 3, 5, 5, 6, 7, 22, 32, 34, 46, 344, 545])
But this still leaves arr alone:
In [68]: arr
Out[68]: [5, 3, 7, 2, 6, 34, 46, 344, 545, 32, 5, 22]
If you want to do it in place (modify the original), use:
In [69]: arr.sort()
In [70]: arr
Out[70]: [2, 3, 5, 5, 6, 7, 22, 32, 34, 46, 344, 545]
np.argsort doesn't sort the list in place, it returns a list full of indicies that you are able to use to sort the list.
You must assign this returned list to a value:
new_arr = np.argsort(arr)
Then, to sort the list with such indices, you can do:
np.array(arr)[new_arr]
Try
order = np.argsort(arr)
print np.array(arr)[order]
the argsort response is the index of the elements.
If you want your array sorted in-place you want arr.sort():
In [1]: import numpy as np
In [2]: arr = [5,3,7,2,6,34,46,344,545,32,5,22]
In [4]: print arr
[5, 3, 7, 2, 6, 34, 46, 344, 545, 32, 5, 22]
In [5]: arr.sort()
In [7]: print arr
[2, 3, 5, 5, 6, 7, 22, 32, 34, 46, 344, 545]
