Article reprint address (reposting it is also convenient for you to find when you use it later)
https://www.cnblogs.com/liyuxia713/p/7082091.html
If you want to sort by specifying a certain row or column, change it to the first row/last row, or the first column/last column to sort, and then switch it back
import numpy as np
>>> a
array([[ 2, 7, 4, 2],
[35, 9, 1, 5],
[22, 12, 3, 2]])
# 按最后一列顺序排序
>>> a[np.lexsort(a.T)]
array([[22, 12, 3, 2],
[ 2, 7, 4, 2],
[35, 9, 1, 5]])
# 按最后一列逆序排序
>>>a[np.lexsort(-a.T)]
array([[35, 9, 1, 5],
[ 2, 7, 4, 2],
[22, 12, 3, 2]])
# 按第一列顺序排序
>>> a[np.lexsort(a[:,::-1].T)]
array([[ 2, 7, 4, 2],
[22, 12, 3, 2],
[35, 9, 1, 5]])
# 按最后一行顺序排序
>>> a.T[np.lexsort(a)].T
array([[ 2, 4, 7, 2],
[ 5, 1, 9, 35],
[ 2, 3, 12, 22]])
按第一行顺序排序
>>> a.T[np.lexsort(a[::-1,:])].T
array([[ 2, 2, 4, 7],
[ 5, 35, 1, 9],
[ 2, 22, 3, 12]])