2

For example,

datetimeidx = pd.DatetimeIndex(
          ['1999-03-01', '1999-03-02', '1999-03-03', '1999-03-04',
           '1999-03-05', '1999-03-08', '1999-03-09', '1999-03-10',
           '1999-03-11', '1999-03-12', '2021-11-16', '2021-11-17', 
           '2021-11-18', '2021-11-19', '2021-11-22', '2021-11-23', 
           '2021-11-24', '2021-11-26', '2021-11-29', '2021-11-30'])

if n=3, what I want is:

datetimeidx = pd.DatetimeIndex(
          ['1999-03-10','1999-03-11', '1999-03-12', 
           '2021-11-26', '2021-11-29', '2021-11-30'])

The point is that I want to select only from the 'given' pd.DatetimeIndex

FObersteiner
  • 22,500
  • 8
  • 42
  • 72
Yanghoon
  • 572
  • 5
  • 18

2 Answers2

2

you could group by year and month, then use the pandas.Series.tail;

n = 3

pd.DatetimeIndex(datetimeidx
                  .to_series()
                  .groupby([datetimeidx.year, datetimeidx.month])
                  .tail(n))

DatetimeIndex(['1999-03-10', '1999-03-11', '1999-03-12', '2021-11-26',
               '2021-11-29', '2021-11-30'],
              dtype='datetime64[ns]', freq=None)
FObersteiner
  • 22,500
  • 8
  • 42
  • 72
2

You can use:

g = pd.Series(datetimeidx.year).astype(str) + '-' + pd.Series(datetimeidx.month).astype(str)
print(pd.DatetimeIndex(pd.Series(datetimeidx).sort_values().groupby(g).tail(3)))
Muhammad Hassan
  • 4,079
  • 1
  • 13
  • 27