Two query single output

Question

Hi,
I am using 10.2.0.1 on windows xp and need your help please:
create table test (datecol date,col1 number(1),col2 number(1));
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),2,3);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),3,4);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),7,1);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),3,7);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),7,7);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),3,3);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),4,4);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),6,4);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),0,4);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),0,0);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),4,0);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),5,0);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),0,8);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),9,8);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),6,8);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),6,0);
insert into test values (to_date('2010-07-25','yyyy-mm-dd'),9,9);
Required output:
      COL1       GTOT         COL2       GTOT
         0          3          0          4
         2          1          1          1
         3          3          3          2
         4          2          4          4
         5          1          7          2
         6          3          8          3
         7          2          9          1
         9          2Means output of :
select col1,count(*) gtot from test group by col1 order by 1;
and
select col2,count(*) gtot from test group by col2 order by 1;
should be looks like above. And if i want to add another column like datecol of test table
then how do i get another column(s).
Thanks.

Answer 1

Hi,
Interesting problem!
What does each row of output represent?
The 1st row represents the lowest value of each of the columns being summarized.
The 2nd row represents the 2nd lowest value of each of the columns.
The 3rd row represents the 3rd lowest value of each of the columns.
The nth row represents the nth lowest value of each of the columns.
Start by assigning numbers 1, 2, 3, ..., n (let's call this number r) to each of the rows, indicating if this is the lowest, 2nd lowest, 3rd lowest, ..., nth lowest value. DENSE_RANK is perfecct for that.
You could unpivot the data onto a separate row for each of the columns being summarized, and then pivot them back into r rows. The approach below is a variant on that, in effect unpivoting only the r value. This way, the values being summarized do not all have to be of the same data type; you can easily have some NUMBER columns and a DATE column, for example.
WITH     got_rs     AS
     SELECT     col1, col2
     ,     DENSE_RANK () OVER (ORDER BY col1)     AS col1_r
     ,     DENSE_RANK () OVER (ORDER BY col2)     AS col2_r
     FROM     test
,     all_rs     AS
     SELECT     col1_r     AS r
     FROM     got_rs
    UNION
     SELECT     col2_r     AS r
     FROM     got_rs
SELECT       MIN   (CASE WHEN g.col1_r = a.r THEN col1 END)     AS col1
, NULLIF (COUNT (CASE WHEN g.col1_r = a.r THEN 1    END), 0)     AS col1_tot
,       MIN   (CASE WHEN g.col2_r = a.r THEN col2 END)     AS col2
, NULLIF (COUNT (CASE WHEN g.col2_r = a.r THEN 1    END), 0)     AS col2_tot
FROM            all_rs     a
LEFT OUTER JOIN     got_rs     g  ON  a.r IN (col1_r, col2_r)
GROUP BY  a.r
ORDER BY  a.r
;