SQL Server joining two tables counting [closed] - sql

Question

Closed. This question needs details or clarity. It is not currently accepting answers.
Want to improve this question? Add details and clarify the problem by editing this post.
Closed 9 years ago.
Improve this question
I want to join table 1 and table 2 by counting the people in table two who visited the combo in table 1, so the result would look like
table 1
City1 City2
Rome Berlin
Moscow Berlin
....many more rows (to be clear- this was just an excerpt)
table 2
Person City
Peter Rome
Peter Paris
Peter Berlin
Peter Moscow
Paul Rome
Paul Moscow
Mary London
Mary Moscow
Mary Berlin
... many more rows
What I want it to look like:
Rome Berlin 0
Moscow Berlin 3
London Berlin 1
Any help appreciated.
Here is what I have done so far:
select a.city1
, a.city2
, count(case when city=a.city1 and city=a.city2 THEN city ELSE NULL END)
from table1 a join table2 b

Answer 1

One way you can do this is by joining to each of the cities and then aggregating, counting up when there is a match in both tables. This form of the query assumes there are no duplicates in the second table:
select t1.city1, t2.city2,
sum(case when t21.person is not null and t22.person is not null then 1 else 0 end) as cnt
from table1 t1 left outer join
table2 t21
on t1.city1 = t21.city left outer join
table2 t22
on t1.city1 = t22.city and t22.person = t21.person
group by t1.city1, t2.city2;
EDIT:
You can also do this with two levels of aggregation and one join. This is pretty easy to explain. The inner query counts the number of cities that a person matches for each pair of cities. The cnt is 0, 1, or 2. The outer query then counts the number where cnt = 2, because these are the ones that match both cities.
select city1, city2, sum(case when cnt = 2 then 1 else 0 end) as cnt
from (select t1.city1, t2.city2, person, count(t2.person) as cnt
from table1 t1 left outer join
table2 t2
on t2.city in (t1.city1, t1.city2);
The advantage of this approach is that adding new cities is easy, as well as getting more information, such as how many people match exactly one of the cities.

Answer 2

Sample data is incorrect, no London-Berlin combination in table1 and result should be
Rome Berlin 1
Moscow Berlin 2
London Berlin 1
However it can be achieved: you want the combinations of the cities of table2 so join table2 to itself based on person.
You also want to mach the combinations in table1 so join table1 to the previous result.
You need a counter so group on the last resultset:
select t2.city, t3.city, count(t1.city1) as counter
from table2 as t2
inner join table2 as t3 on t2.person = t3.person and t2.city <> t3.city
inner join table1 as t1 on t1.city1 = t2.city and t1.city2 = t3.city
group by t2.city, t3.city
The result will be as requested but you have to enter correct data in table1, otherwise you can get wrong result.