objectif

Créer des tableaux de records typés (structured arrays).

code minimal

import numpy as np

dt = np.dtype([("id","i4"), ("x","f4")])
a = np.array([(1, 1.5), (2, 2.5)], dtype=dt)
print(a["id"].tolist())

utilisation

import numpy as np

a = np.zeros(3, dtype=[("r","f4"),("g","f4"),("b","f4")])
a["r"] = [0.1,0.2,0.3]
print(a["r"].tolist())

variante(s) utile(s)

import numpy as np

dt = np.dtype([("k","U3"),("v","i4")])
print(np.array([("abc",1)], dtype=dt)["k"][0])

notes

• Moins flexible que pandas; utile pour IO binaire.