CODEWITHSUNDEEP
pythonarraysnumpy
Kyungmin Kim
Kyungmin Kim

Reputation: 69

Understanding the shape of a numpy.array

If the array x is declared as:

x = np.array([[1, 2], [3, 4]])

the shape of x is (2, 2) because it is a 2x2 matrix.

However, for a 1-dimensional vector, such as:

x = np.array([1, 2, 3])

why does the shape of x gives (3,) and not (1,3)?

Is it my mistake to understand the shape as (row, column)?

Upvotes: 4

Views: 3732

Answers (3)

MisterMiyagi
MisterMiyagi

Reputation: 52139

np.array represents an n-dimensional array. This may include a 2-dimensional array to represent a matrix, for which (row, column) is appropriate. It may also include 1-dimensional, 3-dimensional or other arrays for which (row, column) are too many/few dimensions. Compare:

>>> # 1-dimensional
>>> np.array([1, 2, 3]).shape
(3,)
>>> np.array([1, 2, 3])[1]
2
>>> # 2-dimensional
>>> np.array([[1, 2, 3]]).shape
(1, 3)
>>> np.array([[1, 2, 3]])[0,1]
2
>>> np.array([[1], [2], [3]]).shape
(3, 1)
>>> np.array([[1], [2], [3]])[1, 0]
2
>>> # 3-dimensional
>>> np.array([[[1, 2, 3]]]).shape
(1, 1, 3)
>>> np.array([[[1, 2, 3]]])[0,0,1]
2
>>> np.array([[[1,2],[3,4]],[[5, 6], [7, 8]]]).shape
(2, 2, 2)

Note how the shapes (3,), (1, 3), (3, 1), (1, 1, 3), ... represent different logical layouts, as exemplified by the different position at which a specific element resides.

Upvotes: 1

Pierre D
Pierre D

Reputation: 26271

As per the docs, np.array's are multidimensional containers. Consider:

np.array([[[0, 1], [2, 3]], [[4, 5], [6, 7]]]).shape
# (2, 2, 2)

Also, np.shape always returns a tuple. In your example, (3,) is the representation of a one-element tuple, which is the correct value for the shape of a one-dimensional array.

Upvotes: 1

Leon Markes
Leon Markes

Reputation: 442

Because np.array([1,2,3]) is one-dimensional array. (3,) means that this is single dimension with three elements.

(1,3) means that this is a two-dimensional array. If you use reshape() method on the array, and give it arguments (1,3), additional brackets will be added to it.

>>> np.array([1,2,3]).reshape(1,3)
array([[1, 2, 3]])

Upvotes: 5

Related Questions