Data types ¶

ndarray.dtype : the data type of an array.

One crucial difference in ndarray versus lists is that a ndarray can only contain one data type. Elements must be all floats, ints, strings, bools, etc. This is because NumPy is heavily focused on performance, and enforcing this rule let's them design blazing fast algorithms.

What are all of the data types?

Note

All of these data types are attributes of NumPy. For example, to use int8 you could specify "int8" as dtype or np.int8 .

Integers ¶

Type	Alias	Description
int8	i1	An 8-bit signed integer whose values exist on the interval [-128, +127] .
int16	i2	A 16-bit signed integer whose values exist on the interval [−32,767, +32,767] .
int32	i4	A 32-bit signed integer whose values exist on the interval [−2,147,483,647, +2,147,483,647] .
int64	i8	A 64-bit signed integer whose values exist on the interval [−9,223,372,036,854,775,807, +9,223,372,036,854,775,807] .

Type	Alias	Description
uint8	u1	An 8-bit unsigned integer whose values exist on the interval [0, +255] .
uint16	u2	A 16-bit unsigned integer whose values exist on the interval [0, +65,535] .
uint32	u4	A 32-bit unsigned integer whose values exist on the interval [0, +4,294,967,295] .
uint64	u8	A 64-bit unsigned integer whose values exist on the interval [0, +18,446,744,073,709,551,615] .

Floats ¶

We almost always use float64 .

Type	Alias	Description	Min > 0	Max > 0
float32	f4	IEEE 754 single-precision binary floating-point number.	1.18 ⨉ 10 -38	3.40 ⨉ 10 38
float64	f8	IEEE 754 double-precision binary floating-point number.	2.23 ⨉ 10 -308	1.80 ⨉ 10 308

The table below shows the absolute precision for both formats over a range of values given by

$$ \text{precision} = \frac{\text{value}}{2^{n}}; $$

where $n$ is the number of mantissa bits which is 23 for float32 and 52 for float64 . This can be used to select an appropriate format given the expected value of a number and the required precision.

Value	float32 precision	float64 precision
10 -12	10 -19	10 -28
10 -9	10 -16	10 -25
10 -6	10 -13	10 -22
10 -3	10 -10	10 -19
10 0	10 -7	10 -16
10 3	10 -4	10 -13
10 6	10 -1	10 -10
10 9	10 2	10 -7
10 12	10 5	10 -4

If you are primarily focused on speed, then check to see if the amount of precision loss is negligible.

Text ¶

We generally use Unicode with type utf8 .

Boolean ¶

Type bool is used for True ( 1 ) and False ( 0 ).