连接具有相同列的两个熊猫数据框并合并具有相同索引的行

Question

我有两个数据框 df1 和 df2，每个都有相同的 列名，使用时间戳作为索引。我想连接两个数据帧，同时合并具有相同索引的行，选择存储在 df2 中的值作为首选项。这措辞不好，但见下文。 例如

>>> df1= TimeStamp A_Output B_Output C_Output
          00:00:00  20       15       5
          00:00:06  20       NaN      3
          00:00:15  15       6      NaN
          00:00:20  20       NaN      5
          00:00:30  25       14      10


 >>> df2= TimeStamp A_Output B_Output C_Output
          00:00:00  15       5        8
          00:00:04  16       NaN      NaN
          00:00:06  17       NaN      NaN
          00:00:15  NaN      NaN      2
          00:00:18  19       NaN      NaN
          00:00:21  14       NaN      NaN
          00:00:26  32       NaN      5
          

 >>> df3= TimeStamp A_Output B_Output C_Output
          00:00:00  15       5        8
          00:00:04  16       NaN      NaN
          00:00:06  17       NaN      3
          00:00:15  15       6        2
          00:00:18  19       NaN      NaN
          00:00:21  14       NaN      NaN
          00:00:26  32       NaN      5
          00:00:30  25       14      10

df3 是我想要实现的。 df1 和 df2 中的每个索引都有一个时间戳。对于每个公共索引，其中 db2 不是 NaN，我们取值，否则我们保留存储在 df1 中的值。

df1 >>> 00:00:15  15        6     NaN
df2 >>> 00:00:15  NaN      NaN     2
df3 >>> 00:00:15  15        6      2

df1 >>> 00:00:00  20        15     5
df2 >>> 00:00:00  15         5     8
df3 >>> 00:00:00  15         5     8

有关说明，请参见上述示例。 我真的找不到这样做的方法——作为参考，每个数据帧有大约 90 列和 100k+ 行。

Answer 1

先尝试合并：

df3 = df2.combine_first(df1)

print(df3)

           A_Output  B_Output  C_Output
TimeStamp                              
00:00:00       15.0       5.0       8.0
00:00:04       16.0       NaN       NaN
00:00:06       17.0       NaN       3.0
00:00:15       15.0       6.0       2.0
00:00:18       19.0       NaN       NaN
00:00:20       20.0       NaN       5.0
00:00:21       14.0       NaN       NaN
00:00:26       32.0       NaN       5.0
00:00:30       25.0      14.0      10.0