CODEWITHSUNDEEP
pythonnumpy
M.E.
M.E.

Reputation: 5496

How do I convert a list of tuples into a structured numpy array?

I have the following list:

[(10, 1.1, 1.2, 12),
 (11, 2.1, 2.2, 22),
 (12, 3.1, 3.2, 32),
 (13, 4.1, 4.2, 42)]

And the following numpy datatype:

dt=np.dtype([('a', np.uint32), ('b', np.float64), ('c', np.float64), ('d', np.uint32)])

How could I convert above list into a numpy array of 4 rows 1 column having each row the elements stored as a numpy structured datatype?

Upvotes: 2

Views: 290

Answers (1)

Grismar
Grismar

Reputation: 31319

User @hpaulj already answered in the comments, but with a bit more detail:

import numpy as np

data = [
    (10, 1.1, 1.2, 12),
    (11, 2.1, 2.2, 22),
    (12, 3.1, 3.2, 32),
    (13, 4.1, 4.2, 42)
]

dt = np.dtype([('a', np.uint32), ('b', np.float64), ('c', np.float64), ('d', np.uint32)])

arr = np.array(data, dt)

print(arr)
print(arr.dtype)
print(arr['a'], arr['b'])
print(arr[0], arr[1])

Output:

[(10, 1.1, 1.2, 12) (11, 2.1, 2.2, 22) (12, 3.1, 3.2, 32)
 (13, 4.1, 4.2, 42)]
[('a', '<u4'), ('b', '<f8'), ('c', '<f8'), ('d', '<u4')]
[10 11 12 13] [1.1 2.1 3.1 4.1]
(10, 1.1, 1.2, 12) (11, 2.1, 2.2, 22)

Note that the shape of the arr array is (4,), but considering that's just an array of 4 elements, each element a named structured datatype as required, I would assume it meets the need in the cleanest way.

If somehow the solution doesn't meet your needs, please post a comment indicating what you need to be different and why.

Upvotes: 1

Related Questions