Nested subquery, using items from other part of query

Question

Hi All,
I have a tricky problem and I will try explain as best I can:
DB version is 11.2.0.3.0
create table product_list (product_id number,
                project_name varchar2(30))
create table products (product_id number,
                project_desc varchar2(30))
create table total_counts (product_id number,
                total_count number,
                project_name varchar2(30))
create table fixed_counts (product_id number,
                fixed_count number,
                project_name varchar2(30).
                    fixed_date date)
create table product_rating (product_id number,
                rating number,
                quarter_last_updated number)
create table quarters (quarter_id number,
                quarter_end_date date)
insert into product_list values (1, 'Prod 1');
insert into product_list values (2, 'Prod 2');
insert into product_list values (3, 'Prod 3');
insert into products values (1, 'Prod 1');
insert into products values (2, 'Prod 2');
insert into products values (3, 'Prod 3');
insert into total_counts values (1, 2000, 'Prod 1');
insert into total_counts values (2, 1000, 'Prod 2');
insert into total_counts values (3, 500, 'Prod 3');
insert into fixed_counts values (1, 1, 'Prod 1', to_date('01/01/2013','DD/MM/YYYY'));
insert into fixed_counts values (1, 3, 'Prod 1', to_date('01/01/2013','DD/MM/YYYY'));
insert into fixed_counts values (2, 50, 'Prod 2', to_date('01/01/2013','DD/MM/YYYY'));
insert into fixed_counts values (2, 2, 'Prod 2', to_date('01/01/2013','DD/MM/YYYY'));
insert into fixed_counts values (2, 3, 'Prod 2', to_date('01/01/2013','DD/MM/YYYY'));
insert into fixed_counts values (2, 3, 'Prod 2', to_date('01/01/2013','DD/MM/YYYY'));
insert into fixed_counts values (3, 8, 'Prod 3', to_date('01/01/2013','DD/MM/YYYY'));
insert into fixed_counts values (3, 3, 'Prod 3', to_date('01/03/2013','DD/MM/YYYY'));
insert into product_rating values (1, 1, 1);
insert into product_rating values (3, 2, 2);
insert into quarters values (1, to_date('01/10/2012','DD/MM/YYYY'));
insert into quarters values (2, to_date('01/02/2013','DD/MM/YYYY'));My current query effectively joins 2 tables 'TOTAL_COUNTS' and 'FIXED_COUNTS'. The FIXED_COUNTS table is updated daily when one of the products has a fix made against it and the query outputs a list of all Products and their percentage "fixed" rate (used in an APEX Interactive report).
select A.PRODUCT, A.TOTAL, B.FIXED, a.PROD_ID, round((FIXED/TOTAL) * 100,  2) percent
from (
select sum(a.total_count) TOTAL, a.product_id PROD_ID, d.Product_desc PRODUCT
from total_counts a, product_list c, products d
where a.product_id = c.product_id
and lower(a.project_name) = lower(c.project_name)
and c.product_id = d.product_id
group by a.product_id, d.product_desc
) A
JOIN
select sum(b.fixed_count) FIXED, b.product_id PROD_ID, d.Product_desc PRODUCT
from fixed_counts b, product_list c, products d
where b.product_id = c.product_id
and lower(b.project_name) = lower(c.project_name)
and c.product_id = d.product_id
group by b.product_id, d.product_desc
) B
on A.PROD_ID = B.PROD_ID and A.PRODUCT = B.PRODUCT This gives me an output like:
PRODUCT     TOTAL     FIXED     PROD_ID     percent
Prod 1     2000     4     1     0.2
Prod 2     1000     58     2     5.8
Prod 2      500     11     3     2.2The query works fine and does exactly the job I require. However, I now have a requirement that when a product gets a "fixed" percentage rate over 1% the product gets a point added against it in a "product_rating" table and i have to recalculate the fixed percentage from the date the point was added (if no point added we take all "fixed").
I have added 2 new tables for this: "product_rating" and "quarters". The "product_rating" table contains the Product_id and the points it has accumulated so far. It is updated once a quarter. The "quarters" table just holds the dates for us to use for the new calculations.
So effectively I need to replace this:
sum(b.fixed_count) FIXEDwith something like this:
if a product has a point against it in the "product_rating" table then we just show fixes from the date the point was added i.e.
select sum(b.fixed_count) from fixed_counts b, product_rating c, quarters d
where b.product_id = c.product_id and c.quarter_last_updated = d.quarter_id
and b.product_id = PROD_ID
and b.fixed_date > (select quarter_end_date from quarters where quarter_id = the_last_quarter_updated)Based on the values in the tables above the percentages for 'Prod 3' should now be 0.6. This is because it received a product_rating on quarter 2 so the new calculation is based on all "fixes" after this date (only the last one on 01-MAR-13').
I have tried numerous ways of embedding this in the current query but I cannot get it to work. Sorry for not adding the complete table info but the total_counts and fixed_counts tables have much more columns that I removed from the query so it would be easier to view. Thanks in advance for your help. If I can add anymore info please ask.
Tom
Edited by: TomH on May 14, 2013 8:54 AM
Edited by: TomH on May 14, 2013 8:58 AM

Answer 1

TomH wrote:
of course youre right, thanks.
I have added some table info and data that will hopefully make it a bit easier to understand.Okay, thanks ... in the future though please try and run the commands yourself to ensure they work. What you provided is close, but it didn't run straight away without some modifications.
I'll start with expressing my concerns for your data model, it seems .... wrong, for lack of a better word. Even given that you said you stripped out a bunch of stuff to make the example. Do you have anyone you work with that can review the model for you?
As for the query, this isn't pretty (I'm in a bit of a rush here) so you can probably clean it up considerably. I just tried to demonstrate the basics of what you're going to have to do.
ME_XE? with base_data as
  2  (
  3     select
  4        b.fixed_date,
  5        b.fixed_count,
  6        b.product_id ,
  7        d.project_desc
  8     from fixed_counts b, product_list c, products d
  9     where b.product_id = c.product_id
10     and lower(b.project_name) = lower(c.project_name)
11     and c.product_id = d.product_id
12  ),
13     final_base_data as
14  (
15     select
16        b.product_id      PROD_ID,
17        b.project_desc    PRODUCT,
18        sum
19        (
20           case
21              when b.fixed_date >= stuff.quarter_end_date or stuff.quarter_end_date is null
22              then
23                 b.fixed_count
24              else
25                 0
26           end
27        )  as FIXED
28     from base_data b
29     left outer join
30     (
31        select
32           pr.product_id,
33           qr.quarter_end_date
34        from product_rating pr , quarters qr
35        where pr.quarter_last_updated = qr.quarter_id
36     )  stuff
37     on (stuff.product_id = b.product_id)
38     group by b.product_id, b.project_desc
39  )
40  select A.PRODUCT, A.TOTAL, B.FIXED, a.PROD_ID, round((b.FIXED/a.TOTAL) * 100,  2) percent
41  from
42  (
43     select sum(a.total_count) TOTAL, a.product_id PROD_ID, d.project_desc PRODUCT
44     from total_counts a, product_list c, products d
45     where a.product_id = c.product_id
46     and lower(a.project_name) = lower(c.project_name)
47     and c.product_id = d.product_id
48     group by a.product_id, d.project_desc
49  ) A
50  JOIN final_base_data B
51  on A.PROD_ID = B.PROD_ID and A.PRODUCT = B.PRODUCT
52  ;
PRODUCT                                     TOTAL              FIXED            PROD_ID            PERCENT
Prod 1                                       2000                  4                  1                 .2
Prod 2                                       1000                 58                  2                5.8
Prod 3                                        500                  3                  3                 .6
3 rows selected.Cheers,