Extract Rows with highest values in the columns, MAX doesn't work

[sukh]

From the table i want everything highlighted with a *

I wanted an SQl expression to look at values in Column 1 (ID), look at
the corresponding values in the second column (F1) and select the row
with the highest value, and then if there was more than one row for
that ID with the same value in F1, look at column 3 (F2) and select
the row with the highest value in this column.

SQL> select * from test;

ID F1 F2
---------- ---------- ----------
1 12 4
* 1 12 6
1 11 1
1 9 12
2 3 5
* 2 9 13
2 9 9
3 1 2
3 1 1
* 3 7 5

I try the following

SQL> select * from test t1
2 where f1 = (select max(f1) from test t2 where t2.id = t1.id)
3 ;

And get the following

ID F1 F2
---------- ---------- ----------
1 12 4
* 1 12 6
* 2 9 13
2 9 9
* 3 7 5

If I add another line with an AND statement after line 2 I either get
nothing or rubbish as the output.

Ideas?

Cheers

[Harold Helmich]

If I understand correctly, this should get u going in the right direction:

select a.*
from test a
inner join (
select id
, f1
, max(f2) maxf2
from test
group by id
, f1
) b on b.id = a.id
and b.f1 = a.f1
and b.maxf2 = a.f2

Harry Helmich

sukh wrote:

From the table i want everything highlighted with a *

I wanted an SQl expression to look at values in Column 1 (ID), look at
the corresponding values in the second column (F1) and select the row
with the highest value, and then if there was more than one row for
that ID with the same value in F1, look at column 3 (F2) and select
the row with the highest value in this column.

SQL> select * from test;

ID F1 F2
---------- ---------- ----------
1 12 4
* 1 12 6
1 11 1
1 9 12
2 3 5
* 2 9 13
2 9 9
3 1 2
3 1 1
* 3 7 5

I try the following

SQL> select * from test t1
2 where f1 = (select max(f1) from test t2 where t2.id = t1.id)
3 ;

And get the following

ID F1 F2
---------- ---------- ----------
1 12 4
* 1 12 6
* 2 9 13
2 9 9
* 3 7 5

If I add another line with an AND statement after line 2 I either get
nothing or rubbish as the output.

Ideas?

Cheers

[John Gilson]

"sukh" <su**@jatt.com> wrote in message news:e7************************@posting.google.com ...

From the table i want everything highlighted with a *

I wanted an SQl expression to look at values in Column 1 (ID), look at
the corresponding values in the second column (F1) and select the row
with the highest value, and then if there was more than one row for
that ID with the same value in F1, look at column 3 (F2) and select
the row with the highest value in this column.

SQL> select * from test;

ID F1 F2
---------- ---------- ----------
1 12 4
* 1 12 6
1 11 1
1 9 12
2 3 5
* 2 9 13
2 9 9
3 1 2
3 1 1
* 3 7 5

I try the following

SQL> select * from test t1
2 where f1 = (select max(f1) from test t2 where t2.id = t1.id)
3 ;

And get the following

ID F1 F2
---------- ---------- ----------
1 12 4
* 1 12 6
* 2 9 13
2 9 9
* 3 7 5

If I add another line with an AND statement after line 2 I either get
nothing or rubbish as the output.

Ideas?

Cheers

CREATE TABLE Test
(
id INT NOT NULL,
f1 INT NOT NULL,
f2 INT NOT NULL,
PRIMARY KEY (id, f1, f2)
)

INSERT INTO Test (id, f1, f2)
VALUES (1, 12, 4)
INSERT INTO Test (id, f1, f2)
VALUES (1, 12, 6)
INSERT INTO Test (id, f1, f2)
VALUES (1, 11, 1)
INSERT INTO Test (id, f1, f2)
VALUES (1, 9, 12)
INSERT INTO Test (id, f1, f2)
VALUES (2, 3, 5)
INSERT INTO Test (id, f1, f2)
VALUES (2, 9, 13)
INSERT INTO Test (id, f1, f2)
VALUES (2, 9, 9)
INSERT INTO Test (id, f1, f2)
VALUES (3, 1, 2)
INSERT INTO Test (id, f1, f2)
VALUES (3, 1, 1)
INSERT INTO Test (id, f1, f2)
VALUES (3, 7, 5)

SELECT T1.*
FROM Test AS T1
LEFT OUTER JOIN
Test AS T2
ON T1.id = T2.id AND
(T2.f1 > T1.f1 OR (T2.f1 = T1.f1 AND T2.f2 > T1.f2))
WHERE T2.id IS NULL
ORDER BY T1.id

id f1 f2
1 12 6
2 9 13
3 7 5

Regards,
jag