Convert pandas offset to python date

pandas, python

Solution

If need use `to_offset`:

from pandas.tseries.frequencies import to_offset

ts = pd.to_datetime('2017-05-11') - to_offset("10D")
print (ts)
2017-05-01 00:00:00

print (type(ts))
<class 'pandas._libs.tslib.Timestamp'>

For string add `strftime`:

ts_str = ts.strftime('%Y-%m-%d')
print (ts_str)
2017-05-01
print (type(ts_str))
<class 'str'>

And for date add `date()`:

ts_python_date = ts.date()
print (ts_python_date)
2017-05-01
print (type(ts_python_date))
<class 'datetime.date'>

Another solution is use `Timedelta`:

print (pd.to_datetime('2017-05-11') - pd.Timedelta('10D'))
#same as
#print ((pd.to_datetime('2017-05-11') - pd.to_timedelta('10D')))
2017-05-01 00:00:00

Problem

Pandas has the handy method to_offset, in package pandas.tseries.frequency, which converts a string to an offset: ``` from pandas.tseries.frequencies import to_offset _30_days_ago = to_offset("30D") ``` How can I convert from an offset to: - a Python date, or - a string in format `yyyy-mm-dd` In particular, how can I use offset to calculate dates? For example, if today is 2017-05-11, how can I use `to_offset("10D")` to get the date `2017-05-01` ?

Original source