ID, ATTRIBUTE
1, thing2
1, thing3
1, thing3
2, thing7
2, thing7
2, thing2
3, thing1
3, thing2

mode = lambda x: x.mode() if len(x) > 2 else 'multithing'
df_grouped = df.groupby(['ID'], as_index=False)['ATTRIBUTE].agg(mode)

ID, ATTRIBUTE
1, thing3
2, thing7
3, multithing

df_final.groupby('ATTRIBUTE')['ID'].count()