Resample with DateOffset can give NotImplementedError

Question

Docs for resample say it can take a DateOffset as the rule.
But trying this out sometimes results in a NotImplementedError.

• Passing a string e.g. resample('2M') works.
• The equivalent resample(pd.offsets.MonthEnd(2)) also works.
• But the similar (though different) resample(pd.offsets.DateOffset(months=2)) fails.

Here's code to reproduce:

idx = pd.date_range('20190101', periods=14, freq='3W')
s   = pd.Series(range(len(idx)), index=idx)
s.resample(pd.offsets.DateOffset(months=2)).mean()

Trace:

Traceback (most recent call last):
    ....
  File "....core\generic.py", line 8449, in resample
    level=level,
  File "....core\resample.py", line 1305, in resample
    tg = TimeGrouper(**kwds)
  File "....core\resample.py", line 1378, in __init__
    rule = freq.rule_code
  File "....tseries\offsets.py", line 442, in rule_code
    return self._prefix
  File "....tseries\offsets.py", line 438, in _prefix
    raise NotImplementedError("Prefix not defined")
NotImplementedError: Prefix not defined

Am I right that resample supports only certain DateOffsets?

What kinds doesn't/does it support?

Answer 1

As the comments have stated it is a bug but you can work around it:

According to the documentation, the rule= parameter of DataFrame.resample() can be of type DateOffset, Timedelta or str. Apparently, it actually expects some of the DateOffset subclasses (like MonthEnd, BusinessDay, etc.), because rule_code attribute is not defined in the parent class.

You can either use an equivalent TimeDelta or one of the subclasses of DateOffset