问题
I have a pandas.core.series.Series of specific dates (non-consecutive) which looks like the following:
0 1998-06-09
1 1998-08-07
2 1998-09-11
3 1998-10-13
4 1998-11-03
...
231 2019-07-25
232 2019-09-12
233 2019-10-24
These dates are not necessarily business days. What I need is to create a list or series of dates which represent the previous day of the list that I have. In the new list, every previous day must be a business day, i.e. if (let's say) 1998-06-09 was a Monday, in the new list I should get 1998-06-05, i.e. the previous Friday. Any suggestions?
回答1:
pd.offsets.BusinessDay
When you consider a Business Day to be Monday - Friday
df['prev_bd'] = df['Date'] - pd.offsets.BusinessDay(n=1)
Date prev_bd
0 2012-12-20 2012-12-19
1 2012-12-21 2012-12-20
2 2012-12-22 2012-12-21
3 2012-12-23 2012-12-21
4 2012-12-24 2012-12-21
5 2012-12-25 2012-12-24
6 2012-12-26 2012-12-25
7 2012-12-27 2012-12-26
8 2012-12-28 2012-12-27
9 2012-12-29 2012-12-28
pd.offsets.CustomBusinessDay
When you also want to skip holidays (or any set of dates in particular). Here we exclude Federal Holidays, though you can skip whatever dates you want by providing the appropriate calendar.
from pandas.tseries.holiday import USFederalHolidayCalendar
df['prev_nohol'] = df['Date'] - pd.offsets.CustomBusinessDay(n=1, calendar=USFederalHolidayCalendar())
#PerformanceWarning
Date prev_bd prev_nohol
0 2012-12-20 2012-12-19 2012-12-19
1 2012-12-21 2012-12-20 2012-12-20
2 2012-12-22 2012-12-21 2012-12-21
3 2012-12-23 2012-12-21 2012-12-21
4 2012-12-24 2012-12-21 2012-12-21
5 2012-12-25 2012-12-24 2012-12-24
6 2012-12-26 2012-12-25 2012-12-24 # Christmas skipped because Federal Holiday
7 2012-12-27 2012-12-26 2012-12-26
8 2012-12-28 2012-12-27 2012-12-27
9 2012-12-29 2012-12-28 2012-12-28
Sample Data
import pandas as pd
df = pd.DataFrame({'Date': pd.date_range('2012-12-20', periods=10, freq='D')})
来源：https://stackoverflow.com/questions/60117208/previous-business-dates-pandas