NumPy has a much richer set of data types than standard
Python. The type object for a NumPy type is called a
dtype
.
Each type has a type name. This table shows the names and
describes their functions. Seven of these types are
functionally identical to standard Python types, so their
names have an underscore (_
) appended to
differentiate them. The exact meanings may depend on the
underlying hardware platform.
bool_ 
Boolean, True or False .

byte  Onebyte signed integer. 
short 
Clanguage short integer.

intc 
Clanguage int integer.

int_ 
Python int .

int8 , int16 , int32
 Signed integers with 8, 16, and 32 bits respectively. 
longlong  Signed arbitrarylength integer. 
intp  An integer sized to hold a memory address. 
ubyte  Unsigned 8bit integer. 
ushort 
Unsigned C short .

uintc 
Unsigned C int .

uint  Unsigned integer. 
uint8 , uint16 , uint32
 Unsigned integers of 8, 16, and 32 bits respectively. 
ulonglong  Unsigned arbitrarylength integer. 
uintp  Unsigned integer big enough to hold an address. 
single  Singleprecision float. 
float_ 
Python float .

longfloat  Doubleprecision float. 
float32 , float64
 Floats of 32 and 64 bits, respectively. 
csingle  Singleprecision complex. 
complex_ 
Python complex .

clongfloat  Doubleprecision complex. 
object_  For arrays of arbitrary Python objects. 
str_  Eightbit character string. 
unicode_  Thirtytwobit character string. 