>>> df.idxmax(axis=1) 0 Communications 1 Business 2 Communications 3 Communications 4 Business dtype: object