如何强制 Pandas 应用于 return 父数据框的所有列？

Question

在数据框的某些列上使用 groupby 之后，随后使用 apply 测试字符串是否存在于另一列中，pandas 仅 return 分组依据的列和最后一列使用应用程序创建。是否可以 return 与 groupby 关联的所有列并进行测试？例如，按对话线程的唯一标识符分组，并测试字符串是否存在于另一列中，然后包括数据框中存在但属于特定组的其他一些列？

我试过使用 groupby，然后使用申请匿名函数。

df.head()

 shipment_id shipper_id courier_id  Question                                sender
0   14      9962    228898  Let's get your furbabys home Apple pet transpo...   courier
1   91919   190872  196838  Hi I'm kevin thims and I'm happy to do the job...   courier
2   92187   191128  196838  Hi I'm kevin thims and I'm happy to do the job...   shipper

unique_thread_indentifier = ['shipment_id', 'shipper_id', 'courier_id']
required_variables = ['shipment_id', 'shipper_id', 'courier_id', 'Question', 'sender']

df_new = (
    df
    .groupby(unique_thread_indentifier)[required_variables]
    .apply(lambda group: 'shipper' in group['sender'].unique())
    .to_frame(name='shipper_replied')
    .reset_index()
)

df_new.head()
    shipment_id shipper_id  courier_id  shipper_replied
0   14      9962            228898          False
1   91919   190872          196838          False
2   92187   191128          196838          True

我的目标是将列 Question 和 sender 包含在最终数据框中。预期输出如下所示：

 shipment_id shipper_id courier_id  Question                                sender        shipper_replied
0   14      9962    228898  Let's get your furbabys home Apple pet transpo...   courier       False
1   91919   190872  196838  Hi I'm kevin thims and I'm happy to do the job...   courier       False
2   92187   191128  196838  Hi I'm kevin thims and I'm happy to do the job...   shipper       True

Answer 1

我相信你需要GroupBy.transform:

df['shipper_replied'] = (df.groupby(unique_thread_indentifier)['sender']
                           .transform(lambda group: 'shipper' in group.unique()))

print (df)
   shipment_id  shipper_id  courier_id  \
0           14        9962      228898   
1        91919      190872      196838   
2        92187      191128      196838   

                                          Question   sender  shipper_replied  
0  Let's get your furbabys home Apple pet transpo.  courier            False  
1   Hi I'm kevin thims and I'm happy to do the job  courier            False  
2   Hi I'm kevin thims and I'm happy to do the job  shipper             True

另一个解决方案：

df['shipper_replied'] = (df.assign(new = df['sender'].eq('shipper'))
                           .groupby(unique_thread_indentifier)['new']
                           .transform('any'))

print (df)
   shipment_id  shipper_id  courier_id  \
0           14        9962      228898   
1        91919      190872      196838   
2        92187      191128      196838   

                                          Question   sender  shipper_replied  
0  Let's get your furbabys home Apple pet transpo.  courier            False  
1   Hi I'm kevin thims and I'm happy to do the job  courier            False  
2   Hi I'm kevin thims and I'm happy to do the job  shipper             True

如何强制 Pandas 应用于 return 父数据框的所有列？

How to force Pandas apply to return all columns of parent dataframe?

pandas

pandas-apply

pandas-groupby