Sorting the np array by second value

Question

If I want to sort this array which is declared by np.zeros((50),dtype=object) and I want to sort this array by first value,

[list([2, 5]) list([3, 0]) list([2, 7]) list([3, 1]) list([11, 2])]

like

[list([2, 5]) list([2, 7]) list([3, 0]) list([3, 1]) list([11, 2])]

What should I use? I've been try sort(), sorted() etc..

Your question is a duplicate - also, because `[2, 5]` is a string literal, the `list()` in `list([2, 5])` is redundant. — Grismar, Oct 07 '20 at 00:00
Just checking. Is your input supposed to be a 5x2 numpy array? A list of 5 2-element number arrays? It's kind of unclear. — Frank Yellin, Oct 07 '20 at 00:01
Please, remove salutations. [Should 'Hi', 'thanks', taglines, and salutations be removed from posts?](https://meta.stackexchange.com/questions/2950/should-hi-thanks-taglines-and-salutations-be-removed-from-posts) — maciejwww, Oct 07 '20 at 00:16
@Grismar Ok, I'll do some research. This is my code, `ary[i] = [min(t4), np.argmin(t4)]`, so the `dtype` of array is object, that caused the `list()` type in array. — Harry Lin, Oct 07 '20 at 00:23
@FrankYellin The origin array is a 2D array, but I want to add two integer into this array. — Harry Lin, Oct 07 '20 at 00:26
@LinChengHan Please see if the posted solution resolves the issue. — Ehsan, Oct 07 '20 at 00:29

Ehsan · Accepted Answer · 2020-10-07T00:32:40.187

0

if all lists in your array are of the same length, you can use numpy lexsort (a is your array):

a = np.array(a)
a[np.lexsort((a[:,-1],a[:,-2]))]
#[[ 2  5]
# [ 2  7]
# [ 3  0]
# [ 3  1]
# [11  2]]

or equivalently:

a[np.lexsort(a.T[::-1,:])]

edited Oct 07 '20 at 00:32

answered Oct 07 '20 at 00:26

Ehsan

1 Answers1