I need statistics with count operation for more than two tables:
SELECT T1.city,
(select count(*) from T2 where T1.id=T2.id )
as totbooks ,
(select count(*) from T3 where T1.id=T3.id
and descr like '%newspaper%' )
as totnewspaper
FROM T1
group by T1.descriptionThis is similar to a recent question, Number of rows in two tables (24 May, 2007).
The difference here is that the counts have to be counts of related rows. You were quite right to use correlated subqueries to achieve the counts. However, your GROUP BY is both wrong (because it does not contain the non-aggregate city column that's in the SELECT clause) and unnecessary. Here's what your query should look like:
select T1.city
, ( select count(*)
from T2
where id = T1.id ) as totbooks
, ( select count(*)
from T3
where id = T1.id
and descr like '%newspaper%' )
as totnewspaper
from T1Notice that there is no GROUP BY. The subqueries effectively achieve the "grouping" and each returns a single value to the outer query.
Oracle White Papers: Fusion Middleware