what are all the dtypes that pandas recognizes?

EDIT Feb 2020 following pandas 1.0.0 release

Pandas mostly uses NumPy arrays and dtypes for each Series (a dataframe is a collection of Series, each which can have its own dtype). NumPy's documentation further explains dtype, data types, and data type objects. In addition, the answer provided by @lcameron05 provides an excellent description of the numpy dtypes. Furthermore, the pandas docs on dtypes have a lot of additional information.

The main types stored in pandas objects are float, int, bool, datetime64[ns], timedelta[ns], and object. In addition these dtypes have item sizes, e.g. int64 and int32.

By default integer types are int64 and float types are float64, REGARDLESS of platform (32-bit or 64-bit). The following will all result in int64 dtypes.

Numpy, however will choose platform-dependent types when creating arrays. The following WILL result in int32 on 32-bit platform. One of the major changes to version 1.0.0 of pandas is the introduction of pd.NA to represent scalar missing values (rather than the previous values of np.nan, pd.NaT or None, depending on usage).

Pandas extends NumPy's type system and also allows users to write their on extension types. The following lists all of pandas extension types.

1) Time zone handling

Kind of data: tz-aware datetime (note that NumPy does not support timezone-aware datetimes).

Data type: DatetimeTZDtype

Scalar: Timestamp

Array: arrays.DatetimeArray

String Aliases: 'datetime64[ns, ]'

2) Categorical data

Kind of data: Categorical

Data type: CategoricalDtype

Scalar: (none)

Array: Categorical

String Aliases: 'category'

3) Time span representation

Kind of data: period (time spans)

Data type: PeriodDtype

Scalar: Period

Array: arrays.PeriodArray

String Aliases: 'period[]', 'Period[]'

4) Sparse data structures

Kind of data: sparse

Data type: SparseDtype

Scalar: (none)

Array: arrays.SparseArray

String Aliases: 'Sparse', 'Sparse[int]', 'Sparse[float]'

5) IntervalIndex

Kind of data: intervals

Data type: IntervalDtype

Scalar: Interval

Array: arrays.IntervalArray

String Aliases: 'interval', 'Interval', 'Interval[<numpy_dtype>]', 'Interval[datetime64[ns, ]]', 'Interval[timedelta64[]]'

6) Nullable integer data type

Kind of data: nullable integer

Data type: Int64Dtype, ...

Scalar: (none)

Array: arrays.IntegerArray

String Aliases: 'Int8', 'Int16', 'Int32', 'Int64', 'UInt8', 'UInt16', 'UInt32', 'UInt64'

7) Working with text data

Kind of data: Strings

Data type: StringDtype

Scalar: str

Array: arrays.StringArray

String Aliases: 'string'

8) Boolean data with missing values

Kind of data: Boolean (with NA)

Data type: BooleanDtype

Scalar: bool

Array: arrays.BooleanArray

String Aliases: 'boolean'

pandas borrows its dtypes from numpy. For demonstration of this see the following:

import pandas as pd

df = pd.DataFrame({'A': [1,'C',2.]})
df['A'].dtype

>>> dtype('O')

type(df['A'].dtype)

>>> numpy.dtype

You can find the list of valid numpy.dtypes in the documentation:

'?' boolean

'b' (signed) byte

'B' unsigned byte

'i' (signed) integer

'u' unsigned integer

'f' floating-point

'c' complex-floating point

'm' timedelta

'M' datetime

'O' (Python) objects

'S', 'a' zero-terminated bytes (not recommended)

'U' Unicode string

'V' raw data (void)

pandas should support these types. Using the astype method of a pandas.Series object with any of the above options as the input argument will result in pandas trying to convert the Series to that type (or at the very least falling back to object type); 'u' is the only one that I see pandas not understanding at all:

df['A'].astype('u')

>>> TypeError: data type "u" not understood

This is a numpy error that results because the 'u' needs to be followed by a number specifying the number of bytes per item in (which needs to be valid):

import numpy as np

np.dtype('u')

>>> TypeError: data type "u" not understood

np.dtype('u1')

>>> dtype('uint8')

np.dtype('u2')

>>> dtype('uint16')

np.dtype('u4')

>>> dtype('uint32')

np.dtype('u8')

>>> dtype('uint64')

# testing another invalid argument
np.dtype('u3')

>>> TypeError: data type "u3" not understood

To summarise, the astype methods of pandas objects will try and do something sensible with any argument that is valid for numpy.dtype. Note that numpy.dtype('f') is the same as numpy.dtype('float32') and numpy.dtype('f8') is the same as numpy.dtype('float64') etc. Same goes for passing the arguments to pandas astype methods.

To locate the respective data type classes in NumPy, the Pandas docs recommends this:

def subdtypes(dtype):
    subs = dtype.__subclasses__()
    if not subs:
        return dtype
    return [dtype, [subdtypes(dt) for dt in subs]]

subdtypes(np.generic)

Output:

[numpy.generic,
 [[numpy.number,
   [[numpy.integer,
     [[numpy.signedinteger,
       [numpy.int8,
        numpy.int16,
        numpy.int32,
        numpy.int64,
        numpy.int64,
        numpy.timedelta64]],
      [numpy.unsignedinteger,
       [numpy.uint8,
        numpy.uint16,
        numpy.uint32,
        numpy.uint64,
        numpy.uint64]]]],
    [numpy.inexact,
     [[numpy.floating,
       [numpy.float16, numpy.float32, numpy.float64, numpy.float128]],
      [numpy.complexfloating,
       [numpy.complex64, numpy.complex128, numpy.complex256]]]]]],
  [numpy.flexible,
   [[numpy.character, [numpy.bytes_, numpy.str_]],
    [numpy.void, [numpy.record]]]],
  numpy.bool_,
  numpy.datetime64,
  numpy.object_]]

Pandas accepts these classes as valid types. For example, dtype={'A': np.float}.

NumPy docs contain more details and a chart: