How to check if a pandas Series contains Timestamps?

Question

Following seems to work at least in some cases:

series.dtype == np.dtype('<M8[ns]')

but it doesn't look pretty and I'm not sure if it works always (with all(?) kinds of Timestamps).

Is there a nicer way for testing whether a column contains Timestamps?

Answer 1

try this it should work:

pd.core.dtypes.common.is_datetime_or_timedelta_dtype(series)

if you go through pd.core.dtypes.common.is_ you will find many options for checking timestamps. If you want to combine them then you can use logical operators like:

pd.core.dtypes.common.is_datetime64_ns_dtype(ser)|pd.core.dtypes.common.is_timedelta64_ns_dtype(ser)

Answer 2

The approach in @shivsn's answer is simpler and most likely better. This is a harder way that nonetheless may be informative.

See here for more details on dtypes. But in a nutshell, the second character of the dtype string should be 'M' for a Datetime and 'm' for a Timedelta. Hence, you could test dtype.str[1] == 'M' to detect a Datetime only or dtype.str[1].lower() == 'm' to detect either a Datetime or Timedelta.

>>> dr = pd.date_range('1-1-2017',periods=4,freq='d')
>>> df = pd.DataFrame({ 'i':range(3), 'x':[1.1,2.2,3.3], 
                        'ts':dr[:3], 'td':dr[1:]-dr[:3] })

   i     td         ts    x
0  0 1 days 2017-01-01  1.1
1  1 1 days 2017-01-02  2.2
2  2 1 days 2017-01-03  3.3

>>> for v in df.columns:  
        print( '\ncolumn ' + v + ': ')
        print( 'dtype.str: ', df[v].dtype.str )
        print( 'dtype:     ', df[v].dtype )
        print( 'timestamp? ', df[v].dtype.str[1] == 'M' )

column i: 
dtype.str:  <i8
dtype:      int64
timestamp?  False

column td: 
dtype.str:  <m8[ns]
dtype:      timedelta64[ns]
timestamp?  False

column ts: 
dtype.str:  <M8[ns]
dtype:      datetime64[ns]
timestamp?  True

column x: 
dtype.str:  <f8
dtype:      float64
timestamp?  False