How can I create a numpy array with shape (0, 2)?

Question

I'm trying to understand how a np.array can be created with shape(0,2).

Is this even possible?

So

np.array([]).shape
#output (0,)

np.array([[],[]]).shape
#output (2, 0)

can I get (0, 2) ?

(without using .T)

The problem with nested lists is that the outer list will necessarily contain lists, and won't have a 0 size. Shape (0,2) means 0 'rows', 2 'columns'. — hpaulj
– hpaulj, Commented Feb 5, 2019 at 17:48

busybear · Accepted Answer · 2019-02-05 17:44:21Z

11

Use a function such as np.zeros. No need to worry about the value it generates since there are no values in a 0 x 2 array:

arr = np.zeros((0, 2))
arr.shape

(0, 2)

answered Feb 5, 2019 at 17:44

busybear

10.7k1 gold badge29 silver badges44 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Note that the display of arr array([], shape=(0, 2), dtype=float64) does not use any nested brackets. It has an explicit shape instead.

Safak Ozdek · Accepted Answer · 2019-02-05 18:51:15Z

2

It can be done by np.empty((0,2)) or np.zeros((0,2)).

np.empty(), unlike np.zeros(), does not set the array values to zero and it is slightly faster.

Extra information: In your code np.array([]).shape outputs a rank-1 array which is equal to its transpose.

answered Feb 5, 2019 at 18:45

Safak Ozdek

1,01512 silver badges18 bronze badges