What does ',' do in this slicing operation - [:,::-1]

Question

While going through Sklearn Tutorial I came accross this piece of code

order_centroids = original_space_centroids.argsort()[:, ::-1]

I do not understand what :, does.

The data stored in order_centroids is

array([[28060, 36086, 36087, ..., 29380, 28915, 28914],
       [28060, 33378, 33379, ...,  9698, 26784, 15313],
       [28060, 36209, 36211, ..., 15303, 22350, 48197],
       ...,
       [28060, 36664, 36665, ..., 47821, 32892, 37525],
       [56120, 31887, 31888, ...,  9603, 51250, 30224],
       [56120, 33902, 33903, ..., 20843, 14948, 30316]])

Have you read e.g. https://docs.scipy.org/doc/numpy/reference/arrays.indexing.html? — jonrsharpe, Apr 21 '19 at 18:08
Compare the output of `array([[1,2],[3,4]])`, `array([[1,2],[3,4]])[::-1]` and `array([[1,2],[3,4]])[:,::-1]` — tevemadar, Apr 21 '19 at 18:13

score 2 · Accepted Answer · answered Apr 21 '19 at 18:10

2

The comma separates slicing for each dimension in an array. So [:,::-1] is getting all items from the first dimension and all items from the second dimension but reversing the order on the second dimension with the “step” set to -1.

answered Apr 21 '19 at 18:10

Ethan

1,363
1
6
8

score 0 · Answer 2 · answered Apr 21 '19 at 18:08

It's just a normal comma operator that defines a tuple. Your code

order_centroids = original_space_centroids.argsort()[:, ::-1]

is equivalent to

order_centroids = original_space_centroids.argsort().__getitem__(
    (slice(None, None, None), slice(None, None, -1)
)

What does ',' do in this slicing operation - [:,::-1]

2 Answers2