Add axis option to numpy.random.shuffle

[CyanoKobalamyne]

Currently, numpy.random.shuffle only supports shuffling along the first axis only. Many other Numpy functions support the axis keyword argument to specify the axis along which the respective operation should be applied, it would be great to have such an argument for this function as well.

[subhrm]

I would suggest apply_along_axis . PFB an example :

import numpy as np
a = np.arange(40).reshape((5,8))       
# shuffle along axis 1     
np.apply_along_axis(np.random.shuffle,1,a) 
# shuffle along axis 0
np.random.shuffle(a)
print(a)

Example output:

[[28 24 30 27 31 25 26 29]
 [39 36 37 33 38 35 34 32]
 [ 9 15 12  8 14 13 11 10]
 [22 17 23 19 20 16 18 21]
 [ 2  1  5  4  6  0  7  3]]

[WarrenWeckesser]

FYI: #5173

[wkschwartz]

Can this be closed now because of #13829?

[CyanoKobalamyne]

I believe so. Thank you!