NaTType does not support .time()

[nathholf]

I am currently writing a script to take data from an API and save it to an Excel file for non-technical users who need to access the data. One of the requirements is to have separate date and time column, rather than a single datetime column.

I am using pandas 0.17.0.

import pandas as pd

# creating datetime series
dates = pd.date_range('2015-01-01 00:00', '2015-01-01 03:00', freq='H')
dates = pd.Series(dates)
# setting one record to NaT
dates.iloc[0] = pd.NaT
# converting to date 
dates.apply(lambda x: x.date()) # ignores the NaT
# converting to time
dates.apply(lambda x: x.time()) # fails with error "NaTType does not support time"

I expected NaTType to behave in the same way for converting into a time as it did for a date i.e. returning NaT.

The full error is:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-98-b34fccfc4c06> in <module>()
      9 dates.apply(lambda x: x.date()) # ignores the NaT
     10 # converting to time
---> 11 dates.apply(lambda x: x.time()) # fails with error "NaTType does not support time"

C:\Users\Nathan\Anaconda\lib\site-packages\pandas\core\series.pyc in apply(self, func, convert_dtype, args, **kwds)
   2158             values = lib.map_infer(values, lib.Timestamp)
   2159 
-> 2160         mapped = lib.map_infer(values, f, convert=convert_dtype)
   2161         if len(mapped) and isinstance(mapped[0], Series):
   2162             from pandas.core.frame import DataFrame

pandas\src\inference.pyx in pandas.lib.map_infer (pandas\lib.c:62187)()

[jreback]

We had this debate here: #10372

time is a special animal and should be an obvious error and not glossed over if it returned np.nan (as does .date) on a NaTType.

Further using .apply is very inefficient. You can catch the exception and return np.nan if you'd like, or simply use the methods designed for this.

In [55]: dates
Out[55]: 
0                   NaT
1   2015-01-01 01:00:00
2   2015-01-01 02:00:00
3   2015-01-01 03:00:00
dtype: datetime64[ns]

In [56]: dates.dt.date
Out[56]: 
0           NaN
1    2015-01-01
2    2015-01-01
3    2015-01-01
dtype: object

In [57]: dates.dt.time
Out[57]: 
0         NaN
1    01:00:00
2    02:00:00
3    03:00:00
dtype: object

[nathholf]

Thanks @jreback this makes sense.