Pandas 中的数据透视表：日期范围内的 aggfunc sum

Question

I have a pandas dataframe like the following:

#               Date      Name  RG
#-----------------------------------
#      1: 2013-04-25     NameA   1
#      2: 2013-04-25     NameB   3
#      3: 2013-04-25     NameC   1
#      4: 2013-04-25     NameD   2
#      5: 2013-04-25     NameE   1
#     ---                                                                              
#  13379: 2020-02-13     NameA   3
#  13380: 2020-02-13     NameB   1
#  13381: 2020-02-13     NameC   4
#  13382: 2020-02-13     NameD   1
#  13383: 2020-02-13     NameE   1

I want to pivot the table and use Name column as index. Each Date appears now as an individual column, so that, for each Name index, RG is summed during the past six months, eg, RG value for NameA in 2020-02-06 is obtained by adding all RG values for NameA between 2019-08-07 and 2020-02-06. For example:

#          Name     2013-04-25      2013-04-31      2013-05-07   ---   2020-02-06      2020-02-13
#--------------------------------------------------------------------------------------------------
#      1: NameA     1               2               3                  7               23
#      2: NameB     3               3               6                  15              21
#      3: NameC     1               4               5                  16              24
#      4: NameD     2               2               7                  19              40
#      5: NameE     1               4               9                  15              21
#     ---                                                                              
#    276: NameDE    3               4               6                  15              22
#    277: NameDF    1               4               6                  17              22
#    278: NameDG    4               8               9                  11              23
#    279: NameDH    2               3               5                  19              24
#    280: NameDI    1               4               6                  18              20

I could pivot the table by using the following:

df.pivot_table(
    values='RG', index='Name', columns='Date',
    fill_value=0, aggfunc='sum')

However, values in each column should be the cumulative sum of all values during the previous six months for the same Name . How can I modify aggfunc to accomplish such goal?

Answer 1

I found the answer myself. Before pivoting, it is necessary to generate the cumsum for the selected period by following procedure:

for index, row in df.iterrows():
    currentDate = row['Date']
    previousDate = row['Date'] - pd.DateOffset(months=12)
    name = row['Name']
    mask = (df['Date'] > previousDate) & (df['Date'] <= currentDate) & (df['Name'] == name)
    s = df.loc[mask]['RG'].sum()
    df4.loc[index,'RG'] = s