Merge Select Columns Dataframe Columns Into a Multi-Index

Question

I have N dataframes, in this case lets use 2 dfs as an example:

df1 = pd.DataFrame([['a', 2], ['b', 4]], columns=['foo', 'bar'])
df2 = pd.DataFrame([['a', 3], ['b', 5]], columns=['foo', 'bar'])

Which produce:

    foo bar
0   a   2
1   b   4

    foo bar
0   a   3
1   b   5

How can I concat or merge them into a multi-index, where the new column level’s name is based on some external variable attached to the dfs, Eg I will use the df name as an example here:

        df1 df2
    foo bar bar
0   a   2   3
1   b   4   5

The dataframes are guaranteed to have the same foo values in the same order.

Asked By: ateymour

||

Source

Answer 1

You can use a custom concat:

dfs = [df1, df2]

out = (pd.concat({f'df{i}': d.set_index('foo')
                  for i, d in enumerate(dfs, start=1)}, axis=1)
         .reset_index()
      )

Output:

  foo df1 df2
      bar bar
0   a   2   3
1   b   4   5

If really you need the "foo" at the bottom:

out = (pd.concat({f'df{i}': d.set_index('foo')
                  for i, d in enumerate(dfs, start=1)}, axis=1)
         .swaplevel(axis=1)
         .reset_index().swaplevel(axis=1)
      )

Output:

      df1 df2
  foo bar bar
0   a   2   3
1   b   4   5

Answered By: mozway

Merge Select Columns Dataframe Columns Into a Multi-Index

Question:

Answers: