pandas - 将列添加到日期时间驱动的 Groupby 函数

Question

我有一个数据框；

index   UoW Category Description    Date    Channel Trans
ADATE                   
2018-12-31  1603    Pay Infringement    31/12/2018  AustPost    209
2018-12-31  1604    Pay Infringement    31/12/2018  AustPost    14
2019-12-31  1605    Pay Infringement    31/12/2018  CSC         234
2019-12-31  1606    Pay Infringement    31/12/2018  CSC         1
2019-12-31  1607    Pay Infringement    31/12/2018  DTMR Other  1
2018-12-31  1608    Pay Infringement    31/12/2018  Internet    496
2018-12-30  1609    Pay Infringement    30/12/2018  CSC         266

我想在按“ channel ”和“年份”分组后添加列 df['MonthofYear']。

以下内容为我提供了所需的结果，无需额外的列

df['Trans'].groupby([df['Channel'],  df.index.year]).agg(['max', 'min'])

我尝试过:

 df['MonthofYear']=df['Trans'].groupby([df['Channel'],  df.index.year]).agg(['max', 'min']).transform(df.index.month)

希望得到帮助

Answer 1

使用DataFrameGroupBy.idxmax和 DataFrameGroupBy.idxmin对于索引中的日期时间，按 Trans 列的最大值和最小值，然后将值转换为月份:

tup = [('MaxVal','max'),
       ('MinVal', 'min'), 
       ('MonthofYearMin', 'idxmin'),
       ('MonthofYearMax', 'idxmax')]
df1 = df.groupby(['Channel',  df.index.year.rename('year')])['Trans'].agg(tup)

df1['MonthofYearMax'] = df1['MonthofYearMax'].dt.month
df1['MonthofYearMin'] = df1['MonthofYearMin'].dt.month
print (df1)