python – Pandas根据多个匹配的列值合并2个数据帧

第一个数据帧df1：

seq  id                a1        a2   
12  209981             None    None
12  209982            Funds    None
13  209983      Free_Income    None
13  209984      Free_Income    None
14  209985      Free_Income  Hybrid

和我的第二个数据帧df2：

   seq              a1     p1    p2     
   12              Funds  5.71  1.09  
   12        Free_Income  2.18  3.17  
   12             Hybrid  2.88  3.70
   13        Free_Income  2.53  2.64  
   13              Funds  7.08  3.09 
   13             Hybrid  7.28  3.99  
   14        Free_Income  4.53  2.25  
   14             Hybrid  1.89  2.45  
   14              Funds  1.13  2.35  

现在我想要以下格式的输出

seq  id                a1          a2    p1    p2   p3   p4
12  209981             None      None   None  None  None None 
12  209982            Funds      None   5.71  1.09  None None 
13  209983      Free_Income      None   2.53  2.64  None None
13  209984      Free_Income      None   2.53  2.64  None None
14  209985      Free_Income    Hybrid   4.53  2.25  1.89 2.45

映射是

df1.seq = df2.seq

df1.a1 = df2.a1

df1.a2 = df2.a1

解决方法:

你想合并两次.第一次合并的重点是左侧数据帧中的a1和右侧数据帧中的a1.第二个合并您将焦点从左侧数据帧切换到a2.

df1.merge(
    df2,
    left_on=['seq', 'a1'],
    right_on=['seq', 'a1'],
    how='left'
).join(
    df1.merge(
        df2,
        left_on=['seq', 'a2'],
        right_on=['seq', 'a1'],
        how='left'
    )[['p1', 'p2']].rename(columns=dict(p1='p3', p2='p4'))
)

   seq      id           a1      a2    p1    p2    p3    p4
0   12  209981         None    None   NaN   NaN   NaN   NaN
1   12  209982        Funds    None  5.71  1.09   NaN   NaN
2   13  209983  Free_Income    None  2.53  2.64   NaN   NaN
3   13  209984  Free_Income    None  2.53  2.64   NaN   NaN
4   14  209985  Free_Income  Hybrid  4.53  2.25  1.89  2.45