按一列对2D NumPy数组排序

Question

I though this would be super easy but I am struggling a little. I have a data structure as follows

array([[  5.        ,   3.40166205],
   [ 10.        ,   2.72778882],
   [ 15.        ,   2.31881804],
   [ 20.        ,   2.50643777],
   [  1.        ,   3.94076063],
   [  2.        ,   3.80598599],
   [  3.        ,   3.67121134],
   [  6.        ,   3.2668874 ],
   [  7.        ,   3.13211276],
   [  8.        ,   2.99733811],
   [  9.        ,   2.86256347],
   [ 11.        ,   2.64599467],
   [ 12.        ,   2.56420051],
   [ 13.        ,   2.48240635],
   [ 14.        ,   2.4006122 ],
   [ 16.        ,   1.8280531 ],
   [ 17.        ,   1.74625894],
   [ 18.        ,   1.66446479],
   [ 19.        ,   1.58267063],
   [ 20.        ,   1.50087647]])

And I want to sort it ONLY on the first column ... so it is ordered as follows:

array([[1.  ,  3.9], 
       [2.  ,  3.8], 
       ...  , 
       [20. ,  1.5]])

np.sort doesn't seem to work as it moves array to a flat structure. I've also used itemgetter

from operator import itemgetter
sorted(data, key=itemgetter(1))

But this doesn't give me the output I'm looking for.

Help appreciated!

Answer 1

This is a common numpy idiom. You can use argsort (on the first column) + numpy indexing here -

x[x[:, 0].argsort()]

array([[  1.        ,   3.94076063],
       [  2.        ,   3.80598599],
       [  3.        ,   3.67121134],
       [  5.        ,   3.40166205],
       [  6.        ,   3.2668874 ],
       [  7.        ,   3.13211276],
       [  8.        ,   2.99733811],
       [  9.        ,   2.86256347],
       [ 10.        ,   2.72778882],
       [ 11.        ,   2.64599467],
       [ 12.        ,   2.56420051],
       [ 13.        ,   2.48240635],
       [ 14.        ,   2.4006122 ],
       [ 15.        ,   2.31881804],
       [ 16.        ,   1.8280531 ],
       [ 17.        ,   1.74625894],
       [ 18.        ,   1.66446479],
       [ 19.        ,   1.58267063],
       [ 20.        ,   2.50643777],
       [ 20.        ,   1.50087647]])