如何压缩添加pandas具有不同列的Dataframe

Question

I would like to add two dfs in a zip-like manner:

df1:

        X
a   b
1   1   2
1   2   3

df2:

   X
c
1   1
2   2

desired result:

df1+df2=
           X
a   b   c
1   1   1   3
1   1   2   4
1   2   1   4
1   2   2   5

The only idea I have going row by row but that's hideous..

Answer 1

The problem, as it is now, can be solved with broadcasting:

# new values
new_vals = df1.X.values[:,None] + df2.X.values[None,:]

# new dataframe:
new_df = pd.DataFrame(new_vals, index=df1.index, columns=df2.index)

# stack for the multi-index:
new_df.stack()

output:

a  b  c
1  1  1    3
      2    4
   2  1    4
      2    5
dtype: int64

It still works if you have more than one columns, but needs little tweaking on the new_df's columns:

df1 = (pd.DataFrame({'a':[1,1],
                    'b':[1,2],
                    'X':[0,3],
                    'Y':[1,2]})
         .set_index(['a','b'])
      )

df2 = (pd.DataFrame({'c':[1,2,3],
                    'X':[1,2,3],
                    'Y':[0,1,5]})
         .set_index('c')
      )

new_vals = df1.values[:,None] + df2.values[None,:]

new_df = pd.DataFrame(data=new_vals.reshape(len(df1), df2.shape[1]*df2.shape[0]),
                      index=df1.index, 
                      columns=pd.MultiIndex.from_product((df2.index, df2.columns) )
                     )

Output:

       X  Y
a b        
1 1 1  1  1
    2  2  2
    3  3  6
  2 1  4  2
    2  5  3
    3  6  7

Answer 2

It is easy using concat

pd.concat([df1+df2.loc[x] for x in df2.index],1,keys=df2.index).stack(0)
Out[267]: 
       X
a b c   
1 1 1  3
    2  4
  2 1  4
    2  5

Answer 3

Another solution, creating a new MultiIndex.from_tuples from a list comprehension, then using DataFrame.reindex and DataFrame.add :

new_idx = pd.MultiIndex.from_tuples([x + (y,) for x in df1.index.to_flat_index()
                                     for y in df2.index], names=['a', 'b', 'c'])

df1.reindex(new_idx).add(df2)

[out]

       X
a b c   
1 1 1  3
    2  4
  2 1  4
    2  5