Reducing number of columns in case query

Question

Hi all,
I am operating on 10G R/2.
I have a query which produces the columns with the number of counts.
SUM (CASE WHEN LTRIM(RTRIM(W_O)) ='ROKOLBIN' THEN 1 ELSE 0 END) AS "O_ROKOLBIN",
SUM (CASE WHEN LTRIM(RTRIM(W_I)) = 'ROKOLBIN'THEN 1 ELSE 0 END) AS "I_ROKOLBIN",
sum(case when LTRIM(RTRIM(W_O)) ='GULURU' then 1 else 0 end) as "O_GULURU",
sum(case when LTRIM(RTRIM(W_I)) ='GULURU' then 1 else 0 end) as "I_GULURU",
sum(case when LTRIM(RTRIM(W_ON ))='PURE' then 1 else 0 end) as "O_PURE",
I would like to reduce teh number of columns to 2,
One column is for Division.Division can be either O_ROKOLBIN,I_ROKOLBIN,O_GULURU,I_GULURU,O_PURE
The other column is Subtotal.Subtotal are teh sums as above ,SUM (CASE WHEN LTRIM(RTRIM(W_O)) ='ROKOLBIN' THEN 1 .
Can anyone advie how can I acheiev this.
Thanks.

Answer 1

Hi,
CrackerJack wrote:
Hi all,
Thank you for your wonderful ideas,
I took it onboard and redesigned the query to add the dates.
<pre>
SELECT
DATE_DIVISION week_start_date ,
next_day ( DATE_DIVISION - 7, 'Sunday' ) + 6 week_end_date ,
to_number(to_char(DATE_DIVISION,'WW') )week_no ,
'I_'||W_I AS DIVISION ,
COUNT(W_I) CNT
FROM TEST A
</pre>I get an error when I run that. Don't you need to say TEST_A, with an underscore?
<pre>
GROUP BY DATE_DIVISION, next_day ( DATE_DIVISION - 7, 'Sunday' ) + 6, to_number(to_char(DATE_DIVISION,'WW') ), 'I_'||W_I
</pre>
At the moment query is coming up like this:
12-Jan-10     14-Jan-10     2     I_GULURU     1
12-Jan-10     14-Jan-10     2     I_GULURU     1
12-Jan-10     14-Jan-10     2     I_PURE     1
12-Jan-10     14-Jan-10     2     I_PURE     1
12-Jan-10     14-Jan-10     2     I_GULURU     1
12-Jan-10     14-Jan-10     2     I_GULURU     2
12-Jan-10     14-Jan-10     2     I_GULURU     1
12-Jan-10     14-Jan-10     2     I_ROKOLBIN     1
12-Jan-10     14-Jan-10     2     I_GULURU     1
12-Jan-10     14-Jan-10     2     I_ROKOLBIN     1
12-Jan-10     14-Jan-10     2     I_PURE     1
12-Jan-10     14-Jan-10     2     I_GULURU     1I don't think you're getting this output from the sample data you posted. The dates in the sample data are all in February. It would really help if you posted the actual data, and the results you want from that data. In this case, I suspect the whole problem is that the dates in your real data (unlike your sample data) all have different times. If that's so, use TRUNC (date_division), instead of date_division, like this:
SELECT    TRUNC (date_division)     AS week_start_date
,       next_day ( TRUNC (date_division) - 1
             , 'Saturday'
                )           AS week_end_date
,       TO_CHAR ( date_division
            , 'FMWW'
            )          AS week_no
,       'I_' || w_i           AS division
,       COUNT (w_i)           AS cnt
FROM        test_a
GROUP BY  TRUNC (date_division)
,       TO_CHAR ( date_division
            , 'FMWW'
,       w_i
;