Grouping - aggregation issue- Sample data included

Question

The below data shows that this particular course A16 at a top level (set=01) equals to 60% coursework and 40% exam. At this level all the set=01's should always equal to 100%.
The complication is where a piece of Coursework and/or Exam is made up of other parts. So as we can see in the data below, the 40% coursework is made up of an Assessment paper (58%) and a Practical piece (42%). You can see the association between the 01 level, 02, 03 level through the set, subset relationship.
So record with subset 2 has two pieces associated to it.
with t as (              
select 'A16' course, '01' sett, '03' Subset, 'E' Code, 'Exam'         Descr, 40 "weight%"  from dual UNION
  select 'A16' course, '01' sett, '02' Subset, 'C' Code, 'Courswork Total' Descr, 60 "weight%"  from dual UNION
   select 'A16' course, '02' sett, ''   Subset, 'C' Code, '1. Assignement'      Descr, 58 "weight%"  from dual UNION
    select 'A16' course, '02' sett, ''   Subset, 'P' Code, '2. Practical'         Descr, 42 "weight%"  from dual UNION
     select 'A16' course, '03' sett, ''   Subset, 'E' Code, '1. Exam'             Descr, 50 "weight%"  from dual UNION
      select 'A16' course, '03' sett, ''   Subset, 'W' Code, '2. Written Piece'     Descr, 50 "weight%"  from dual)  
select * from t;This is what I had so far but this only looks at the top level which as you can see is no good as it doesnt knoe about the practical elements.
SELECT course
           ,sett
           ,NVL(SUM(CASE WHEN CODE IN ('C','F','J','L','R','Y')          THEN "weight%" end),0) AS Coursework
           ,NVL(SUM(CASE WHEN CODE IN ('E','Q')                          THEN "weight%" end),0) AS Written
           ,NVL(SUM(CASE WHEN CODE IN ('A','D','O','P','S','T','V','W')  THEN "weight%" end),0) AS Practical            
     FROM t
     where sett = 01
   GROUP BY course, sett
   ORDER BY sett; What I am trying to calculated is a total Exam%, Written%, Practical% which when all Summed equal to 100%
EXPECTED Results for the supplied data set are below:
select 'A16' course, 20 Exam,  45.2 Practical, 34.8 Coursework, 20+45.2+34.8 Total from dual;The t.Code relates to whether the piece is coursework, exam or practical. As seen below. So this is how I know what sections relate to which part.
I need to basically sieve through each level and calculate its % of the 01 level and group them into Exam, Practical and Courswork.
           CODE IN ('C','F','J','L','R','Y')          Coursework
           CODE IN ('E','Q')                          Written
           CODE IN ('A','D','O','P','S','T','V','W')  Practical  Any ideas would be much appreciated.

Answer 1

Thanks for that sKr.
abhi: The only issue I have with courses such as the below. This course has only has 01's so it cant roll them up, unless we use the 00 record. But if I change the code to start with 00, it will work for this one. But not the OP data i sent (This is further down this post)
START WITH T.SETT = '00'
with t as (  
--select 'A16' course, '00' sett, '01' Subset, '' Code, 'Generated'  Descr, 100 "WEIGHT"  from dual UNION -- This record is the master record.
select 'A16' course, '01' sett, '' Subset, 'O' Code, 'Presen'  Descr, 10 "WEIGHT"  from dual UNION           
select 'A16' course, '01' sett, '' Subset, 'R' Code, 'Case'    Descr, 70 "WEIGHT"  from dual UNION
select 'A16' course, '01' sett, '' Subset, 'O' Code, 'Poster'  Descr, 10 "WEIGHT"  from dual UNION
select 'A16' course, '01' sett, '' Subset, 'C' Code, 'Journel' Descr, 10 "WEIGHT"  from dual)
SELECT TT.COURSE,
             NVL(SUM(CASE WHEN TT.CODE IN ('E', 'Q') THEN "WW" END), 0)                               AS EXAM,
             NVL(SUM(CASE WHEN TT.CODE IN ('A', 'D', 'O', 'P', 'S', 'T', 'V', 'W') THEN "WW" END), 0) AS PRACTICAL,
             NVL(SUM(CASE WHEN TT.CODE IN ('C', 'F', 'J', 'L', 'R', 'Y') THEN "WW" END),     0)       AS COURSEWORK,
             NVL(SUM(WW),0) TOTAL    
    FROM (SELECT T.*, LEVEL, (CONNECT_BY_ROOT "WEIGHT") * T.WEIGHT / 100 WW
                    FROM T
                 START WITH T.SETT = '01'
                CONNECT BY SETT = PRIOR SUBSET) TT
WHERE TT.SUBSET IS NULL
GROUP BY COURSE;In the OP, the data set had master levels at 01. But there is actual one more level above that which is 00 and all courses have one. I thought I would mention it in-case it can be use to fix the above examples: OP data below with the missing 00 record.
with t as (  
select 'A16' course, '00' sett, '01' Subset, '' Code, 'Generated'         Descr, 100 "WEIGHT"  from dual UNION           
select 'A16' course, '01' sett, '03' Subset, 'E' Code, 'Exam'         Descr, 40 "WEIGHT"  from dual UNION
select 'A16' course, '01' sett, '02' Subset, 'C' Code, 'Courswork Total' Descr, 60 "WEIGHT"  from dual UNION
select 'A16' course, '02' sett, ''   Subset, 'C' Code, '1. Assignement'      Descr, 58 "WEIGHT"  from dual UNION
select 'A16' course, '02' sett, ''   Subset, 'P' Code, '2. Practical'         Descr, 42 "WEIGHT"  from dual UNION
select 'A16' course, '03' sett, ''   Subset, 'E' Code, '1. Exam'             Descr, 50 "WEIGHT"  from dual UNION
select 'A16' course, '03' sett, ''   Subset, 'W' Code, '2. Written Piece'     Descr, 50 "WEIGHT"  from dual)
SELECT TT.COURSE,
             NVL(SUM(CASE WHEN TT.CODE IN ('E', 'Q') THEN "WW" END), 0)                               AS EXAM,
             NVL(SUM(CASE WHEN TT.CODE IN ('A', 'D', 'O', 'P', 'S', 'T', 'V', 'W') THEN "WW" END), 0) AS PRACTICAL,
             NVL(SUM(CASE WHEN TT.CODE IN ('C', 'F', 'J', 'L', 'R', 'Y') THEN "WW" END),     0)       AS COURSEWORK,
             NVL(SUM(WW),0) TOTAL    
    FROM (SELECT T.*, LEVEL, (CONNECT_BY_ROOT "WEIGHT") * T.WEIGHT / 100 WW
                    FROM T
                 START WITH T.SETT = '01'
                CONNECT BY SETT = PRIOR SUBSET) TT
WHERE TT.SUBSET IS NULL
GROUP BY COURSE;
Edited by: oraCraft on Mar 12, 2012 10:16 AM