为每个客户机添加缺少的行- Python / Pandas

Question

我有几个星期，几个月，几年的df。

week = ['01/03/2022 - 01/09/2022', '01/10/2022 - 01/16/2022', '01/17/2022 - 01/23/2022', '01/24/2022 - 01/30/2022']
month = ["January", "January", "January", "January"]
year = [2022, 2022, 2022, 2022]

myDict = {}

myDict["Week"] = week
myDict["Month"] = month 
myDict["Year"] = year

dates_df = pd.DataFrame(data=myDict)
dates_df

         Week              Month    Year

01/03/2022 - 01/09/2022   January   2022
01/10/2022 - 01/16/2022   January   2022
01/17/2022 - 01/23/2022   January   2022
01/24/2022 - 01/30/2022   January   2022

数据看起来类似于(以前分组)：

test_data = {'CLient Id': [1,1,1,1,2,2,2,3,3],
    'Client Name': ['Tom Holland', 'Tom Holland', 'Tom Holland', 'Tom Holland', 'Brad Pitt', 'Brad Pitt', 'Brad Pitt', 'Anna Delvey', 'Anna Delvey'],
    'City': ['New York', 'New York', 'New York', 'New York', 'Los Angeles', 'Los Angeles', 'Los Angeles', 'New York', 'New York'],    
    'Week': ['01/03/2022 - 01/09/2022', '01/10/2022 - 01/16/2022', '01/17/2022 - 01/23/2022', '01/24/2022 - 01/30/2022',
            '01/03/2022 - 01/09/2022', '01/10/2022 - 01/16/2022', '01/24/2022 - 01/30/2022', '01/03/2022 - 01/09/2022',
            '01/24/2022 - 01/30/2022'], 
    'Month': ['January', 'January', 'January', 'January', 'January', 'January', 'January', 'January', 'January'], 
    'Year': [2022, 2022, 2022, 2022, 2022, 2022, 2022, 2022, 2022],
    'Spent': [1000, 750, 1200, 850, 777, 1500, 1200, 1500, 1200]}

test_df = pd.DataFrame(data=test_data)
test_df

Client Id    Client Name          City                Week               Month    Year   Spent

1            Tom Holland       New York      01/03/2022 - 01/09/2022    January   2022   1000
1            Tom Holland       New York      01/10/2022 - 01/16/2022    January   2022   750
1            Tom Holland       New York      01/17/2022 - 01/23/2022    January   2022   1200
1            Tom Holland       New York      01/24/2022 - 01/30/2022    January   2022   850
2            Brad Pitt         Los Angeles   01/03/2022 - 01/09/2022    January   2022   777
2            Brad Pitt         Los Angeles   01/10/2022 - 01/16/2022    January   2022   1500
2            Brad Pitt         Los Angeles   01/24/2022 - 01/30/2022    January   2022   1200
3            Anna Delvey       New York      01/03/2022 - 01/09/2022    January   2022   1500
3            Anna Delvey       New York      01/24/2022 - 01/30/2022    January   2022   1200

我需要为每个缺少一周的客户端(Id)创建额外的客户端名称行，即城市、周、月和年，在本例中，df + be列应该是0。

我需要的产出：

Client Id    Client Name          City                  Week               Month    Year   Spent

1            Tom Holland        New York       01/03/2022 - 01/09/2022    January   2022   1000
1            Tom Holland        New York       01/10/2022 - 01/16/2022    January   2022   750
1            Tom Holland        New York       01/17/2022 - 01/23/2022    January   2022   1200
1            Tom Holland        New York       01/24/2022 - 01/30/2022    January   2022   850
2            Brad Pitt          Los Angeles    01/03/2022 - 01/09/2022    January   2022   777
2            Brad Pitt          Los Angeles    01/10/2022 - 01/16/2022    January   2022   1500
2            Brad Pitt          Los Angeles    01/17/2022 - 01/23/2022    January   2022   0
2            Brad Pitt          Los Angeles    01/24/2022 - 01/30/2022    January   2022   1200
3            Anna Delvey        New York       01/03/2022 - 01/09/2022    January   2022   1500
3            Anna Delvey        New York       01/10/2022 - 01/16/2022    January   2022   0
3            Anna Delvey        New York       01/17/2022 - 01/23/2022    January   2022   0
3            Anna Delvey        New York       01/24/2022 - 01/30/2022    January   2022   1200

我试着对它进行编码，但没有运气。

Answer 1

您可以使用pivot + reindex + fillna (获取丢失的数据)+ stack (返回到以前的形状)：

columns = ['Week','Month','Year']
out = (test_df.pivot(['CLient Id', 'Client Name', 'City'], columns, ['Spent'])
       .reindex(pd.MultiIndex.from_arrays(dates_df.assign(Spent='Spent').to_numpy()[:, [-1,0,1,2]].T, 
                                          names=[None]+columns), axis=1)
       .fillna(0).stack(level=columns).reset_index())

输出：

    CLient Id  Client Name         City                     Week    Month  Year   Spent  
0           1  Tom Holland     New York  01/03/2022 - 01/09/2022  January  2022  1000.0  
1           1  Tom Holland     New York  01/10/2022 - 01/16/2022  January  2022   750.0  
2           1  Tom Holland     New York  01/17/2022 - 01/23/2022  January  2022  1200.0  
3           1  Tom Holland     New York  01/24/2022 - 01/30/2022  January  2022   850.0  
4           2    Brad Pitt  Los Angeles  01/03/2022 - 01/09/2022  January  2022   777.0  
5           2    Brad Pitt  Los Angeles  01/10/2022 - 01/16/2022  January  2022  1500.0  
6           2    Brad Pitt  Los Angeles  01/17/2022 - 01/23/2022  January  2022     0.0  
7           2    Brad Pitt  Los Angeles  01/24/2022 - 01/30/2022  January  2022  1200.0  
8           3  Anna Delvey     New York  01/03/2022 - 01/09/2022  January  2022  1500.0  
9           3  Anna Delvey     New York  01/10/2022 - 01/16/2022  January  2022     0.0  
10          3  Anna Delvey     New York  01/17/2022 - 01/23/2022  January  2022     0.0  
11          3  Anna Delvey     New York  01/24/2022 - 01/30/2022  January  2022  1200.0  

如果有多个列要填充0，则可以使用：

columns = ['Week','Month','Year']
value_columns = ['Spent', ...]
new_df = (pd.DataFrame({'key':[1]*len(value_columns),'New':value_columns})
          .merge(dates_df.assign(key=1)).drop(columns='key'))
out = (test_df.pivot(['CLient Id', 'Client Name', 'City'], columns, value_columns)
       .reindex(pd.MultiIndex.from_arrays(zip(*new_df.to_numpy()), 
                                          names=[None] + columns), axis=1)
       .fillna(0).stack(level=columns).reset_index())