Question

我有一个重复ID的数据框，但数据在多个区域部分完成。

MESSAGE 'Hello' type status to client

我想保留所有数据，以便在数据存在时创建新列，使其看起来像下面的数据框：

我尝试了以下操作，但它删除了我想保留的数据。

df = pd.DataFrame([[1234, 'Customer A', '123 Street', np.nan, np.nan],
               [1234, 'Customer A', np.nan, '333 Street', np.nan],
               [1234, 'Customer A', '12345 Street', np.nan, np.nan],
               [1234, 'Customer A', np.nan, np.nan, np.nan],
               [1233, 'Customer B', '444 Street', '3335 Street', np.nan],
               [1233, 'Customer B', '555 Street', '666 Street', np.nan],
               [1233, 'Customer B', '553 Street', '666 Street', 'abc@email.com'],
               [1235, 'Customer C', '1553 Street', '644 Street', 'abc@email.com'],
               [1235, 'Customer C', '2553 Street', '644 Street', 'abc@email.com']],     
               columns=['ID', 'Customer', 'Billing Address', 'Shipping Address', 'Contact'])


df
        ID  Customer    Billing Address Shipping Address    Contact
0   1234    Customer A  123 Street      NaN                 NaN
1   1234    Customer A  NaN             333 Street          NaN
2   1234    Customer A  12345 Street    NaN                 NaN
3   1234    Customer A  NaN             NaN                 NaN
4   1233    Customer B  444 Street      3335 Street         NaN
5   1233    Customer B  555 Street      666 Street          NaN
6   1233    Customer B  553 Street      666 Street          abc@email.com
7   1235    Customer C  1553 Street     644 Street          abc@email.com
8   1235    Customer C  2553 Street     644 Street          abc@email.com

编辑：我添加了更多数据，因为原始帖子中不清楚可能存在多行ID。

Answer 1

以下是一种使用apply并使用dict创建pd.Series创建新列的方法

In [1057]: cols = ['Billing Address', 'Shipping Address']

In [1058]: (df.groupby(['ID', 'Customer'])
              .apply(lambda g: pd.Series({'%s %s' % (x, i+1): v[x] 
                     for i, v in enumerate(g[cols].to_dict('r'))
                     for x in v})))
Out[1058]:
                Billing Address 1 Billing Address 2 Shipping Address 1  \
ID   Customer
1233 Customer B        444 Street        555 Street         333 Street
1234 Customer A        123 Street               NaN                NaN

                Shipping Address 2
ID   Customer
1233 Customer B         666 Street
1234 Customer A         333 Street

Answer 2

这是一个潜在的解决方案，尽管在该过程中使用的内存方面根本没有效率。

我们的想法是循环使用唯一ID的行数，并将数据框与第n行合并：

new_df = df.drop_duplicates(subset = ['ID'])
temp_df = df.drop(new_df.index)
nth_address = 1
while len(temp_df) > 0:
    temp = temp_df.drop_duplicates(subset = ['ID'])
    new_df = new_df.merge(temp,suffixes = ('_'+str(nth_address),'_'+str(nth_address+1)),\
                          on = 'ID',how = 'left')
    temp_df = temp_df.drop(temp.index)
    nth_address +=1

    ID      Customer_1  Billing Address_1   Shipping Address_1  Customer_2  Billing Address_2   Shipping Address_2
0   1234    Customer A  123 Street          NaN                 Customer A  NaN                 333 Street
1   1233    Customer B  444 Street          333 Street          Customer B  555 Street          666 Street

为了符合您想要的输出，我们需要在['ID','Customer']上合并，因为它在您的示例中是相同的键：

new_df = df.drop_duplicates(subset = ['ID'])
temp_df = df.drop(new_df.index)
nth_address = 1
while len(temp_df) > 0:
    temp = temp_df.drop_duplicates(subset = ['ID'])
    new_df = new_df.merge(temp,suffixes = ('_'+str(nth_address),'_'+str(nth_address+1)),on = ['ID','Customer'],how = 'left')
    temp_df = temp_df.drop(temp.index)
    nth_address+=1

    ID      Customer    Billing Address_1   Shipping Address_1  Billing Address_2   Shipping Address_2
0   1234    Customer A  123 Street          NaN                 NaN                 333 Street
1   1233    Customer B  444 Street          333 Street          555 Street          666 Street

Pandas删除每行中部分完成数据的重复项并组合数据

2 个答案: