Columns of each row in 2D Numpy array do not shuffle

Asked Mar 23 '19 at 16:02

Active Mar 23 '19 at 16:02

Viewed 26 times

I am trying to shuffle each row's columns independently in a Numpy 2D array and I tried using the solution from here, but my array stays the same as before the operation. Here is my code:

perms = np.random.rand(*deltas.shape).argsort(1)    #deltas is a 2d array of float values
deltas = deltas[perms, np.arange(deltas.shape[1])]

As I said deltas stays the same as before after executing deltas = deltas[perms, np.arange(deltas.shape[1])]. I cannot find the problem. I have a Numpy version 1.16.2

EDIT: It seems my indexing is not correct, since it shuffles correctly if I call .argsort(0), but I want it to work with .argsort(1).

asked Mar 23 '19 at 16:02

Luki

3

You need to use `perms` as column indices, so `deltas[np.arange(deltas.shape[0])[:,None], perms]` I think. – Divakar Mar 23 '19 at 16:09
Updated my answer there to shuffle along a generic axis. – Divakar Mar 23 '19 at 16:24

Columns of each row in 2D Numpy array do not shuffle

0 Answers0