pandas使用datetime作为索引并用groupby调用tseries.offsets时间位移分组

import numpy as np
import pandas as pd
from datetime import datetime
from pandas.tseries.offsets import Day, MonthEnd

sj = pd.Series(np.random.randn(20),
               index=pd.date_range('1/15/2022', periods=20, freq='3d'))
sj

2022-01-15    0.852965
2022-01-18   -0.955869
2022-01-21   -0.023493
2022-01-24   -2.304234
2022-01-27   -0.652469
2022-01-30   -1.218302
2022-02-02   -1.332610
2022-02-05    1.074623
2022-02-08    0.723642
2022-02-11    0.690002
2022-02-14    1.001543
2022-02-17   -0.503087
2022-02-20   -0.622274
2022-02-23   -0.921169
2022-02-26   -0.726213
2022-03-01    0.222896
2022-03-04    0.051316
2022-03-07   -1.157719
2022-03-10    0.816707
2022-03-13    0.433610
Freq: 3D, dtype: float64

使用本月末最后一天分组

offset = MonthEnd()
sj.groupby(offset.rollforward).mean()

2022-01-31   -0.716900
2022-02-28   -0.068394
2022-03-31    0.073362
dtype: float64

---

等同于使用重采样

sj.resample('M').mean()