熊猫groupby：串联而不调整大小

Question

I have a pandas data frame with 4 columns:

Col1 Col2 Col3 Col4
A1    B1   C1   X1
A2    B2   C2   X2
A3    B3   C3   X3
A1    B1   C1   X4
A4    B4   C4   X5
A3    B3   C3   X6

I want to identify rows that have same values in col1, col2 and col3 and then concatenate the values in their correspondent col4. So the output would be like:

Col1 Col2 Col3 Col4
A1    B1   C1   X1, X4
A2    B2   C2   X2
A3    B3   C3   X3, X6
A1    B1   C1   X4, X1
A4    B4   C4   X5
A3    B3   C3   X6, X3

The final shape of the data frame is same as the original data frame. It would be great if you someone could point me in the right direction. Thanks

Answer 1

a = (df
     .groupby(['Col1', 'Col2', 'Col3'])['Col4']
     .apply(lambda x: ', '.join(sorted(x)))
    )
b = (df
     .groupby(['Col1', 'Col2', 'Col3'])['Col4']
     .apply(lambda x: ', '.join(sorted(x, reverse=True)))
    )
pd.concat([a, b]).drop_duplicates().reset_index()

And the output:

  Col1 Col2 Col3    Col4
0   A1   B1   C1  X1, X4
1   A2   B2   C2      X2
2   A3   B3   C3  X3, X6
3   A4   B4   C4      X5
4   A1   B1   C1  X4, X1
5   A3   B3   C3  X6, X3

Answer 2

Use transform and not apply or agg.

df['Col4'] = df.groupby(['Col1', 'Col2', 'Col3']).transform(lambda x: ', '.join(x.tolist()))

  Col1 Col2 Col3    Col4
0   A1   B1   C1  X1, X4
1   A2   B2   C2      X2
2   A3   B3   C3  X3, X6
3   A1   B1   C1  X1, X4
4   A4   B4   C4      X5
5   A3   B3   C3  X3, X6