在分組數據框中創建包含兩列對角線組合的新列

Question

我有以下數據框

  ID  Agent Capital_1 Capital_2
0  10      5      Rome     Paris
1  10      5     Paris    Berlin
2  20      6      Rome     Paris
3  20      6     Paris    Madrid

df = pd.DataFrame({'ID':[10,10,20,20],
                   'Agent':[5,5,6,6],
                   'Capital_1':['Rome','Paris','Rome', 'Paris'],
                   'Capital_2':['Paris','Berlin','Paris','Madrid']})

對於ID和Agent的每個組，我想創建新的Capitals列，其中包含兩列Capital_1和Capital_2的對角線組合，如下所示：

預期 Output：

   ID  Agent Capital_1 Capital_2                Capitals
0  10      5      Rome     Paris              RomeBerlin
1  10      5     Paris    Berlin              RomeBerlin
2  20      6      Rome     Paris              RomeMadrid
3  20      6     Paris    Madrid              RomeMadrid

在我的原始數據集中，可能有多個組，這只是一個例子。

是否有捷徑可尋？

Answer 1

我無法將它插入“oneliner”，但它可能會更好，如每個步驟所示：

# grouping by ID/Agent also taking first value from Cap1 and last value from Cap2 (from each group)

df2 = df.groupby(["ID","Agent"]).agg({"Capital_1": "first", "Capital_2": "last"}) 

# creates result, which we can use for merging 
print(df2)
         Capital_1 Capital_2
ID Agent                    
10 5          Rome    Berlin
20 6          Rome    Madrid

# creating the column for Captilas 
df2['Capitals'] = df2['Capital_1']+ df2['Capital_2']

#merging grouped result and old dataframe into the result you need while adding suffix to new duplicated columns
df = df.merge(df2, on=["ID", "Agent"], how="left", suffixes=('', '_Z'))

# getting rid of duplicated colums
df.drop(df.filter(regex='_Z$').columns, axis=1, inplace=True)

print(df)
   ID  Agent Capital_1 Capital_2    Capitals
0  10      5      Rome     Paris  RomeBerlin
1  10      5     Paris    Berlin  RomeBerlin
2  20      6      Rome     Paris  RomeMadrid
3  20      6     Paris    Madrid  RomeMadrid