I tried to use a numpy array with fromiter but It gave this error
import numpy
l=numpy.dtype([("Ad","S20"),("Yas","i4"),("Derecelendirme","f")])
a=numpy.array([("Dr.Wah",20,0.9)])
d=numpy.fromiter(a,dtype=l,count=3)
print(d)
ValueError: setting an array element with a sequence.
In [172]: dt=np.dtype([("Ad","S20"),("Yas","i4"),("Derecelendirme","f")])
...: alist = [("Dr.Wah",20,0.9)]
The normal way to define a structured array is to use a list of tuples for the data along with the dtype:
In [173]: np.array( alist, dtype=dt)
Out[173]:
array([(b'Dr.Wah', 20, 0.9)],
dtype=[('Ad', 'S20'), ('Yas', '<i4'), ('Derecelendirme', '<f4')])
fromiter works as well, but isn't as common
In [174]: np.fromiter( alist, dtype=dt)
Out[174]:
array([(b'Dr.Wah', 20, 0.9)],
dtype=[('Ad', 'S20'), ('Yas', '<i4'), ('Derecelendirme', '<f4')])
If you create an array without the dtype:
In [175]: a = np.array(alist)
In [176]: a
Out[176]: array([['Dr.Wah', '20', '0.9']], dtype='<U6')
In [177]: _.shape
Out[177]: (1, 3)
a.astype(dt) does not work. You have to use a recfunction:
In [179]: import numpy.lib.recfunctions as rf
In [180]: rf.unstructured_to_structured(a, dtype=dt)
Out[180]:
array([(b'Dr.Wah', 20, 0.9)],
dtype=[('Ad', 'S20'), ('Yas', '<i4'), ('Derecelendirme', '<f4')])
awithdtype=l? Otherwise your float values will get coerced to strings.