Pandas DataFrame Groupby two columns and get counts

In [56]: df.groupby(['col5','col2']).size().reset_index().groupby('col2')[[0]].max()
Out[56]: 
      0
col2   
A     3
B     2
C     1
D     3