内部联接与数据库中的两个表

Question

I have 2 tables in sql, and i am doing a query:

in the first query i am counting the data from a date

In the secont query i am counting the data from a the same date

And in the 3 query i am doing a inner join, but i need show the data that i have in the query

20101120    26   19

But i have:

20101120    313 313

I have the example in:

http://sqlfiddle.com/#!4/e61fc/2

Who can help me?

What is the problem?

I can do it with a join?

Or the problem is the join?

Answer 1

SELECT  A.id_date_sus AS id_date,
        A.N1,
        B.N2
FROM (SELECT id_date_sus, 
             COUNT(id_date_sus) AS N1
      FROM suscription
      GROUP BY id_date_sus) A
LEFT JOIN (SELECT id_date, 
                  COUNT(id_date) AS N2
           FROM billing
           GROUP BY id_date) B
  ON A.id_date_sus = B.id_date;

Answer 2

If you use oracle db, You can do something like:

with s1 as
(select id_date_sus, count(id_date_sus) c1
from suscription
group by id_date_sus),

s2 as
(select id_date, count(id_date) c2
from billing
group by id_date)

select s1.id_date_sus, s1.c1, s2.c2 from s1, s2 where s1.id_date_sus = s2.id_date

Answer 3

The problem is your data. There are only two services in each table.

So

 (s.id_service = b.id_service and 
  b.id_date = s.id_date_sus) 

is not joining on a unique key.

When that happens we get a cartesian product, in which every record on one side of the join is paired to every record on the other side of the join. For service_id=1 that produces (19*15) rows; for service_id=2 that produces (4*7) rows. Now, 285+28=313, which explains the count which troubles you.

The solution is to run the two counts as separate sub-queries and then join them in the main query:

select s.id_date_sus, s.cnt as s_cnt, b.cnt as b_cnt 
from 
    ( select s.id_date_sus, count(*) cnt
      from suscription s
      group by s.id_date_sus) s
    join ( select b.id_date, count(*) cnt
           from billing b 
           group by b.id_date) b
    on (s.id_date_sus = b.id_date);

Here is my re-work of your SQL Fiddle .