df.compare() how to change self/other labels with align

If I have two dataframes, then since pandas 1.1.0 I can compare them along axis 1 as follows:

import pandas as pd

df1 = pd.DataFrame([[1,2,3,4], [1,2,3,4]], index=['A', 'B'])
df2 = pd.DataFrame([[1,2,5,4], [5,2,3,1]], index=['A', 'B'])

df1.compare(df2, align_axis=1)

I would like to rename the self/other labels to something more descriptive.

I am aware of

But that doesn't work for align_axis=1:

df1.compare(df2, align_axis=1).rename(index={'self': 'left', 'other': 'right'}, level=-1)

(I also tried with level=0).

I am also aware of this open pull request that would add a suffixes argument to pd.compare that would allow you to do:

df1.compare(df2, align_axis=1, suffixes=["left", "right"])

but until that pull request gets merged, what is the way to currently do this in pandas?

CodePudding user response：

When is axis 1 you should change index to columns in rename

df1.compare(df2, align_axis=1).rename(columns={'self': 'left', 'other': 'right'}, level=-1)
Out[56]: 
     0          2          3      
  left right left right left right
A  NaN   NaN  3.0   5.0  NaN   NaN
B  1.0   5.0  NaN   NaN  4.0   1.0