Question

I have a table XYZ as

+------+------+------+-----+
|  X1  |  x2  |  x3  |  x4 |
+------+------+------+-----+
|  a   |   b  |  c   |  1  |
|  a   |   b  |  d   |  2  |
|  p   |   q  |  e   |  3  |
|  p   |   q  |  f   |  4  |
+------+------+------+-----+

condition is if x1 and x2 matches then get x3 and x4 of
row where x4 is maximum
when i will query this table I want to get output as

+------+------+------+-----+
|  X1  |  x2  |  x3  |  x4 |
+------+------+------+-----+
|  a   |   b  |  d   |  2  |
|  p   |   q  |  f   |  4  |
+------+------+------+-----+

i tried as
select * from XYZ t1,
(select x1,x2,max(x4) from XYZ ) t2
where t1.x1=t2.x1 and t1.x2=t2.x2

but i want to use advance functions like rank and partations
as this is much slow in large database

Answer 1

rank or row_number can be used depending on whether you want to return multiple rows in the case of x4 being equal.

eg:

SQL> select *
  2    from (select x1, x2, x3, x4, rank() over (partition by x1, x2 order by x4 desc) rnk
  3            from xyz)
  4   where rnk = 1;

X X X         X4        RNK
- - - ---------- ----------
a b d          2          1
p q f          4          1

but if for a, b we had two rows at x4=2, then we’d get two rows back:

SQL> insert into xyz values ('a', 'b', 'd', 2);

1 row created.

SQL> select *
  2    from (select x1, x2, x3, x4, rank() over (partition by x1, x2 order by x4 desc) rnk
  3            from xyz)
  4   where rnk = 1;

X X X         X4        RNK
- - - ---------- ----------
a b d          2          1
a b d          2          1
p q f          4          1

if you wanted just one, use row_number() instead:

SQL> select *
  2    from (select x1, x2, x3, x4, row_number() over (partition by x1, x2 order by x4 desc) rn
  3            from xyz)
  4   where rn = 1;

X X X         X4         RN
- - - ---------- ----------
a b d          2          1
p q f          4          1

SQL>