How to convert array from dtype=object to dtype=np.int

Question

Currently, the array I got is

arr = array([array([ 2,  7,  8, 12, 14]), array([ 3,  4,  5,  6,  9, 10]),
   array([0, 1]), array([11, 13])], dtype=object)

How can I convert it into array([[ 2, 7, 8, 12, 14], [ 3, 4, 5, 6, 9, 10], [0, 1], [11, 13]])?

I tried arr.astype(np.int), but failed

You have mixed dimensions, which won't work with numpy. EDIT: well, it will work, but you won't be able to use vectorization. You'd probably be better with lists in that case. — roganjosh
– roganjosh, Commented Sep 2, 2018 at 11:42
Related: How do I stack vectors of different lengths in NumPy? — jpp
– jpp, Commented Sep 2, 2018 at 12:02

jpp · Accepted Answer · 2018-09-02 12:07:43Z

2

The dtype for an array of arrays will always be object. This is unavoidable because with NumPy only non-jagged n-dimensional arrays can be held in a contiguous memory block.

Notice your constituent arrays are already of int dtype:

arr[0].dtype  # dtype('int32')

Notice also your logic will work for a non-jagged array of arrays:

arr = np.array([np.array([ 2,  7,  8]),
                np.array([ 3,  4,  5])], dtype=object)

arr = arr.astype(int)

arr.dtype  # dtype('int32')

In fact, in this case, the array of arrays is collapsed into a single array:

print(arr)

array([[2, 7, 8],
       [3, 4, 5]])

For computations with a jagged array of arrays you may see some performance advantages relative to a list of lists, but the benefit may be limited. See also How do I stack vectors of different lengths in NumPy?

answered Sep 2, 2018 at 12:07

jpp

166k37 gold badges301 silver badges363 bronze badges

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

Collectives™ on Stack Overflow

How to convert array from dtype=object to dtype=np.int

1 Answer 1

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related