Creating Numpy structured array

Question

While brushing up my numpy skills, I've encounter the structured array.

>>> import numpy as np
>>> person_data_def = [('name', 'S6'), ('height', 'f8'), ('weight', 'f8'), ('age', 'i8')]
>>> person_data_def
[('name', 'S6'), ('height', 'f8'), ('weight', 'f8'), ('age', 'i8')]

What are S6, f8 and i8 in this context?

Answer 1

I think you are mixing things up a little bit.

person_data_def is not a structured array but just a list of tuples. However, this list is used as the type for the structured array. Adapting the example from the numpy documentation gives

x = np.array([('Rex', 9, 81.0, 18), ('Fido', 3, 27.0, 20)], dtype=[('name', 'U10'), ('height', 'f8'), ('weight', 'f8'), ('age', 'i8')] )

x can then be accessed by using

x['name'] which will give you array(['Rex', 'Fido'], dtype='<U10')

S6, U10, i8, etc. are the datatypes and a little trickier to understand, but for example i4 refers to a signed integer of size 32 bits (size marked by the number). Also the >and < you will see are to distinguish between little and big endian (direction your computer read bytes). But I think they are set automatically if not specified other way.

See here for a more detailed description of the datatypes: https://numpy.org/doc/stable/reference/arrays.dtypes.html#arrays-dtypes-constructing