Oracle group/count query - oracle

Question

I have the following two tables:
TableOne
========
Id1|ColA1|ColB1|ColC1|ColD1|ColE1
--------------------------------
1| AFoo|BFoo |CFoo | DFoo| EFoo
2| AFoo|BBar |CFoo | DFoo| EFoo
TableTwo
========
Id2|ColA2|ColB2|ColC2
---------------------
11| 1 |ABC |NOP |
12| 1 |ABC |QRS |
13| 1 |DEF |TUV |
14| 1 |DEF |WXY |
15| 1 |DEF |FGH |
16| 2 |ABC |NOP |
I run the following query:
select t1.*, t2.*
from TableOne t1
inner join TableTwo t2 on t2.ColA2=t1.Id1
where t1.ColA1='AFoo'
and get the following result:
Result
======
Id1|ColA1|ColB1|ColC1|ColD1|ColE1|Id2|ColA2|ColB2|ColC2
-------------------------------------------------------
1| AFoo|BFoo |CFoo | DFoo| EFoo| 11| 1 | ABC | NOP
1| AFoo|BFoo |CFoo | DFoo| EFoo| 12| 1 | ABC | QRS
1| AFoo|BFoo |CFoo | DFoo| EFoo| 13| 1 | DEF | TUV
1| AFoo|BFoo |CFoo | DFoo| EFoo| 14| 1 | DEF | WXY
1| AFoo|BFoo |CFoo | DFoo| EFoo| 15| 1 | DEF | FGH
2| AFoo|BBar |CFoo | DFoo| EFoo| 16| 2 | ABC | NOP
What I really want returned is:
Desired Result
======
Id1|MaxDup
----------------------------------------
1| 3 (This is because there are 3 DEF records)
2| 1 (This is because there is 1 ABC record)
So, I am trying to track the maximum number of occurrences in ColB2 which appears for each TableOne row. In the example above, the ID1 of 1 has two ABC records and three DEF records associated with it. Since there are more DEF records than ABC records, I want the count of DEF records returned.
Can anybody provide a working example that can demonstrate this?

Answer 1

Try this one (I haven't tested it):
with t2 as (
select ColA2, ColB2, count(*) cnt
from TableTwo
group by ColA2, ColB2
)
select t1.Id1,
( select max(cnt) MaxDup
from t2
where t2.ColA2=t1.Id1)
from TableOne t1

Answer 2

Here is a result based on your test data.
SQL> select
cola2
, cnt as maxDup
from (
select
cola2
, colb2
, cnt
, rank() over (partition by cola2 order by cnt desc ) cnt_rnk
from (
select cola2
, colb2
, count(*) as cnt
from TableTwo
group by cola2, colb2
)
)
where cnt_rnk = 1
/
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
COLA MAXDUP
---- ----------
1 3
2 1
SQL>
The innermost query counts the duplicate. The middle query ranks the counts of duplicates. The outermost query filters for the combo with the highest count:

Answer 3

Maybe this one?
SELECT t2.id2 id, count(t2.ColB2) MaxDup
FROM TableTwo t2
WHERE EXISTS(
SELECT * FROM TableOne t1
WHERE t1.id1 = t2.id2
AND t1.ColA1='AFoo')
AND rownum < 2
GROUP BY t2.id2, t2.ColB2
ORDER BY MaxDup desc;

Answer 4

It's SQL Server code, but I think it can be safely transferred to Oracle:
select id1,cola1,colb1,colc1,cold1,cole1,max(dup) as dup2 from (
select t1.id1, t1.cola1, t1.colb1, t1.colc1,t1.cold1,t1.cole1, t2.colb2,count(*) as dup
from t1
inner join t2 on t2.cola2=t1.Id1
where t1.cola1='Afoo'
group by t1.id1, t1.cola1, t1.colb1, t1.colc1,t1.cold1,t1.cole1, t2.colb2)
tab
group by id1,cola1,colb1,colc1,cold1,cole1
It works on Sql Server, and by enveloping in a subquery you can extract only id1 and dup2. Tried expanding your dataset, it works.