Try this:
b['col3'] = (b['col1'] + '-' + b['col2'])
print(b.groupby('col3').size())
Output:
a-b 2
b-a 2
c-d 1
d-c 1
EDIT 1
Based on your input data (as in comments), here is the df I made up and the results
Code:
df[['seller_city','customer_city']]
Output:
seller_city customer_city
0 volta redonda sao jose dos pinhais
1 volta redonda sao jose dos pinhais
2 sao jose dos pinhais volta redonda
Code:
df.groupby((df['seller_city'] + '-' + df['customer_city'])).size()
Output:
sao jose dos pinhais-volta redonda 1
volta redonda-sao jose dos pinhais 2