Count by unique pair of columns in pandas [duplicate]

If you use groupby, you will get what you want.

d.groupby(['ip', 'useragent']).size()

produces:

ip          useragent               
192.168.0.1 a           2
            b           1
192.168.0.2 b           1