groupbyhead3 - joingreat/bugList GitHub Wiki df.sort_values(['job','count'],ascending=False).groupby('job').head(3)