pandas (original) (raw)

In [2]: dti = pd.date_range('2016-01-01', periods=3)
In [3]: dti + 1
Out[3]: DatetimeIndex(['2016-01-02', '2016-01-03', '2016-01-04'], dtype='datetime64[ns]', freq='D')
In [4]: dti[1] + 100
Out[4]: Timestamp('2016-04-11 00:00:00', freq='D')

In [5]: tdi = pd.timedelta_range('1D', periods=3)
In [6]: tdi - 1
Out[6]: TimedeltaIndex(['0 days', '1 days', '2 days'], dtype='timedelta64[ns]', freq='D')
In [7]: tdi[-1] - 1
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-7-ac830ffbe976> in <module>()
----> 1 tdi[-1] - 1

TypeError: unsupported operand type(s) for -: 'Timedelta' and 'int'

This TypeError occurs because TimedeltaIndex._box_func does not pass self.freq to Timedelta, and is one of many ways that this behavior introduces inconsistencies into the code.

Bottom line: DTI/TDI/Timestamp/Timedelta arithmetic with integers should be deprecated to a) cut out a bunch of hard-to-maintain code and b) avoid internal inconsistencies.