Help with Oracle Analytic Function scenario

Question

Hi,
I am new to analytic functions and was wondering if someone could help me with the data scenario below. I have a table with the following data
COLUMN A COLUMN B COLUMN C
13368834 34323021 100
13368835 34438258 50
13368834 34438258 50
13368835 34323021 100
The output I want is
COLUMN A COLUMN B COLUMN C
13368834 34323021 100
13368835 34438258 50
A simple DISTINCT won't give me the desired output so i was wondering if there is any way that I can get the result using ANALYTIC FUNCTIONS and DISTINCT ..
Any help will be greatly appreciated.
Thanks.

Answer 1

Hi,
Welcome to the forum!
Whenever you have a question, please post your sample data in a form that people can use to re-create the problem and test their solutions.
For example:
CREATE TABLE     table_x
(      columna     NUMBER
,      columnb     NUMBER
,      columnc     NUMBER
INSERT INTO table_x (columna, columnb, columnc) VALUES (13368834, 34323021, 100);
INSERT INTO table_x (columna, columnb, columnc) VALUES (13368835, 34438258, 50);
INSERT INTO table_x (columna, columnb, columnc) VALUES (13368834, 34438258, 50);
INSERT INTO table_x (columna, columnb, columnc) VALUES (13368835, 34323021, 100);Do you want something that works in your version or Oracle? Of course you do! So tell us which version that is.
How do you get the results that you want? Explain what each row of output represents. It looks like
the 1st row contains the 1st distinct value from each column (where "first" means descending order for columnc, and ascending order for the others),
the 2nd row contains the 2nd distinct value,
the 3rd row contains the 3rd distinct value, and so on.
If that's what you want, here's one way to get it (in Oracle 9 and up):
WITH     got_nums     AS
     SELECT     columna, columnb, columnc
     ,     DENSE_RANK () OVER (ORDER BY  columna        )     AS a_num
     ,     DENSE_RANK () OVER (ORDER BY  columnb        )     AS b_num
     ,     DENSE_RANK () OVER (ORDER BY  columnc  DESC)     AS c_num
     FROM     table_x
SELECT       MAX (a.columna)          AS columna
,       MAX (b.columnb)          AS columnb
,       MAX (c.columnc)          AS columnc
FROM              got_nums     a
FULL OUTER JOIN  got_nums     b     ON     b.b_num     =           a.a_num
FULL OUTER JOIN  got_nums     c     ON     c.c_num     = COALESCE (a.a_num, b.b_num)
GROUP BY  COALESCE (a.a_num, b.b_num, c.c_num)
ORDER BY  COALESCE (a.a_num, b.b_num, c.c_num)
;I've been trying to find a good name for this type of query. The best I've heard so far is "Prix Fixe Query", named after the menus where you get a choice of soups (listed in one column), appetizers (in another column), main dishes (in a 3rd column), and so on. The items on the first row don't necessaily have any relationship to each other.
The solution does not assume that there are the same number of distinct items in each column.
For example, if you add this row to the sample data:
INSERT INTO table_x (columna, columnb, columnc) VALUES (13368835, 34323021, 99);which is a copy of the last row, except that there is a completely new value for columnc, then the output is:
`  COLUMNA    COLUMNB    COLUMNC
  13368834   34323021        100
  13368835   34438258         99
                              50starting in Oracle 11, you can also do this with an unpivot-pivot query.