How to look up identical column names in two dataframes and combine the matched columns

Question

I want to loop trough column names of two data frames, find the columns with identical column name, and combine them to create a new data frame.

I tried to write an if-else condition in a for loop but it doesn't work.

To be specific, I have two data frame like this:

df1 = pd.DataFrame({'A': {'2019Q1': 1, '2019Q2': 2, '2019Q3': 3},
                'B': {'2019Q1': 1, '2019Q2': 3, '2019Q3': 5},
                'C': {'2019Q1': 2, '2019Q2': 4, '2019Q3': 6}})

df2 = pd.DataFrame({'A': {'2019Q1': 4, '2019Q2': 5, '2019Q3': 6},
                'B': {'2019Q1': 1.5, '2019Q2': 3.3, '2019Q3': 5.6},
                'C': {'2019Q1': 2.3, '2019Q2': 4.8, '2019Q3': 6.7}})

I want outputs like below-

for A, output=

pd.DataFrame({'df1': {'2019Q1': 1, '2019Q2': 2, '2019Q3': 3},
              'df2': {'2019Q1': 4, '2019Q2': 5, '2019Q3': 6}) 

for B, output=

pd.DataFrame({'df1': {'2019Q1': 1, '2019Q2': 3, '2019Q3': 5},
              'df2': {'2019Q1': 1.5, '2019Q2': 3.3, '2019Q3': 5.6}) 

for C,output=

pd.DataFrame({'df1': {'2019Q1': 2, '2019Q2': 4, '2019Q3': 6},
              'df2': {'2019Q1': 2.3, '2019Q2': 4.8, '2019Q3': 6.7})

Thank you very much for your help!

Answer 1

Here is one way similar to @ALollz but save the subdf in multiple index dataframe

s = pd.concat([df1, df2], keys=['df1', 'df2']).unstack(0)
s.loc[:,'A']
Out[390]: 
        df1  df2
2019Q1    1    4
2019Q2    2    5
2019Q3    3    6

Answer 2

concat with keys + groupby. Store the results in a dict, with the columns as keys.

d = {idx: gp.droplevel(1, axis=1) for idx, gp in
     pd.concat([df1, df2], keys=['df1', 'df2'], axis=1).groupby(level=1, axis=1)}

d['A']
#        df1  df2
#2019Q1    1    4
#2019Q2    2    5
#2019Q3    3    6

d['B']
#        df1  df2
#2019Q1    1  1.5
#2019Q2    3  3.3
#2019Q3    5  5.6

---

The above will create Frames for all columns regardless if they're found in both. If that's not useful you can change the concat to:

cols = df1.columns.union(df2.columns)
pd.concat([df1[cols], df2[cols]], axis=1, keys=['df1', 'df2'])