Advanced grouping help

Question

Hi,
My tables holds data like this.
Region      Agent       Product     Category   Sale
West     Agent1     P1     C1     2
West     Agent1     P2     C3     3
East     Agent1     P4     C4     2
East     Agent2     P1     C1     5The report we want to generate should look like this
               Region     TotalSales(Total Sales from this region          Top 5 Agents     $     Top 5 Products     $     Top 5 Category     $
               West     200          Agent 1     25     P1     18     C1     23
                              Agent 2     15     P2     15     C2     21
                              Agent 3     15     P3     12     C3     21
                              Agent 4     10     P4     7     C4     20
                              Agent 5     9     P5     6     C5     6
               East     400          Agent 4          P3     75     C5     123
                              Agent 7          P7     30     C8     111Basically everything is pivoted on Region,
Basically what we have is
Columns 1, 2 - sum(sales) group by region
Columns 1, 3, 4 - sum(sales) group by region, agent
Columns 1, 5, 6 - sum(sales) group by region, product
Columns 1, 7, 8 - sum(sales) group by region, category
I'm on 10.2.0.3 and this has to be in SQL (no PL/SQL). I was thinking on using grouping sets but was getting a little confused. I had a temp table using (WITH AS), first grouped by all the columns then grouping by seperately. I was planning use PARTION and rownum to get the top 5 part. I can do this for each individual group by, but all put together seems a little confusing.
So just wanted to put my question out there in case some one has already achieved this.

Answer 1

I don't think so it is so difficult. Is this what you are looking at? First query gives you data based on the sales. The second query gives the region wise sales with agentid, category and product of the top two sales in each region.
SQL> with sample as(
  2  select 1 regionid, 12 agentid, 21 productid, 31 category, 12 sale from dual
  3  union all
  4  select 1, 12, 21, 31, 3 from dual
  5  union all
  6  select 1, 13, 21, 32, 3 from dual
  7  union all
  8  select 1, 13, 22, 31, 5 from dual
  9  union all
10  select 1, 14, 24, 33, 7 from dual
11  union all
12  select 2, 16, 21, 32, 2 from dual
13  union all
14  select 2, 14, 22, 32, 6 from dual
15  union all
16  select 2, 16, 22, 35, 8 from dual
17  union all
18  select 3, 12, 23, 36, 1 from dual
19  union all
20  select 3, 13, 23, 35, 8 from dual
21  union all
22  select 3, 15, 25, 33, 9 from dual
23  union all
24  select 3, 15, 26, 33, 10 from dual
25  ),
26  sample_ranked as
27  (
28  select regionid, sum(sale) over (partition by regionid) tot_sales,
29         row_number() over (partition by regionid order by sale desc) tot_rnk,
30         agentid,
31         productid,
32         category,
33         row_number() over (partition by regionid order by sale desc) rnk,
34         sale
35  from sample
36  )
37  select decode(a.tot_rnk, 1, a.regionid, NULL) regionid,
38         decode(a.tot_rnk, 1, a.tot_sales, NULL) total_sales,
39         a.agentid agentid,
40         a.productid productid,
41         a.category category,
42         a.sale sales
43  from sample_ranked a
44  /
  REGIONID TOTAL_SALES    AGENTID  PRODUCTID   CATEGORY      SALES
         1          30         12         21         31         12
                               14         24         33          7
                               13         22         31          5
                               12         21         31          3
                               13         21         32          3
         2          16         16         22         35          8
                               14         22         32          6
                               16         21         32          2
         3          28         15         26         33         10
                               15         25         33          9
                               13         23         35          8
                               12         23         36          1
12 rows selected.
SQL>
SQL> with sample as(
  2  select 1 regionid, 12 agentid, 21 productid, 31 category, 12 sale from dual
  3  union all
  4  select 1, 12, 21, 31, 3 from dual
  5  union all
  6  select 1, 13, 21, 32, 3 from dual
  7  union all
  8  select 1, 13, 22, 31, 5 from dual
  9  union all
10  select 1, 14, 24, 33, 7 from dual
11  union all
12  select 2, 16, 21, 32, 2 from dual
13  union all
14  select 2, 14, 22, 32, 6 from dual
15  union all
16  select 2, 16, 22, 35, 8 from dual
17  union all
18  select 3, 12, 23, 36, 1 from dual
19  union all
20  select 3, 13, 23, 35, 8 from dual
21  union all
22  select 3, 15, 25, 33, 9 from dual
23  union all
24  select 3, 15, 26, 33, 10 from dual
25  ),
26  sample_ranked as
27  (
28  select regionid, sum(sale) over (partition by regionid) tot_sales,
29         row_number() over (partition by regionid order by sale desc) tot_rnk,
30         agentid,
31         productid,
32         category,
33         row_number() over (partition by regionid order by sale desc) rnk,
34         sale
35  from sample
36  )
37  select decode(a.tot_rnk, 1, a.regionid, NULL) regionid,
38         decode(a.tot_rnk, 1, a.tot_sales, NULL) total_sales,
39         a.agentid agentid,
40         a.productid productid,
41         a.category category,
42         a.sale sales
43  from sample_ranked a
44  where a.rnk <=2
45  /
  REGIONID TOTAL_SALES    AGENTID  PRODUCTID   CATEGORY      SALES
         1          30         12         21         31         12
                               14         24         33          7
         2          16         16         22         35          8
                               14         22         32          6
         3          28         15         26         33         10
                               15         25         33          9
6 rows selected.
SQL> Cheers
Sarma.