How to Retrieve Rank Based on Total Mark in SQLite Table - sql

Question

say I have a table like this:
S_id |ca1 |ca2 |exam
1 | 08 | 12 | 35
1 | 02 | 14 | 32
1 | 08 | 12 | 20
2 | 03 | 11 | 55
2 | 09 | 18 | 45
2 | 10 | 12 | 35
3 | 07 | 12 | 35
3 | 04 | 14 | 37
3 | 09 | 15 | 32
4 | 03 | 11 | 55
4 | 09 | 18 | 45
4 | 10 | 12 | 35
5 | 10 | 12 | 35
5 | 07 | 12 | 35
5 | 09 | 18 | 45
I want to select S_id, total and assign a rank on each student based on sum(ca1+ca2+exam) like the following:
S_id |total|rank
1 | 158 | 5
2 | 198 | 1
3 | 165 | 4
4 | 198 | 1
5 | 183 | 3
If there are the same total, like S_id 2 and S_id 4 with rank 1, I want the rank to be jumped to 3.
Thanks for helping.

Answer 1

Make the table:
sqlite> create table t (S_id, ca1, ca2, exam);
sqlite> insert into t values
...> ( 1 , 08 , 12 , 35 ),
...> ( 1 , 02 , 14 , 32 ),
...> ( 1 , 08 , 12 , 20 ),
...> ( 2 , 03 , 11 , 55 ),
...> ( 2 , 09 , 18 , 45 ),
...> ( 2 , 10 , 12 , 35 ),
...> ( 3 , 07 , 12 , 35 ),
...> ( 3 , 04 , 14 , 37 ),
...> ( 3 , 09 , 15 , 32 ),
...> ( 4 , 03 , 11 , 55 ),
...> ( 4 , 09 , 18 , 45 ),
...> ( 4 , 10 , 12 , 35 ),
...> ( 5 , 10 , 12 , 35 ),
...> ( 5 , 07 , 12 , 35 ),
...> ( 5 , 09 , 18 , 45 );
Make a temporary table with the total scores:
sqlite> create temp table tt
as select S_id, sum(ca1) + sum(ca2) + sum(exam) as total
from t group by S_id;
Use the temporary table to compute the ranks:
sqlite> select s.S_id, s.total,
(select count(*)+1 from tt as r where r.total > s.total) as rank
from tt as s;
1|143|5
2|198|1
3|165|4
4|198|1
5|183|3
Drop the temporary table:
sqlite> drop table tt;
ADDENDUM
With a recent change (2015-02-09) to SQLite, this formulation now works:
with tt (S_id, total) as
(select S_id, sum(ca1 + ca2 + exam) as total from t group by S_id)
select s.S_id, s.total,
(select count(*)+1 from tt as r where r.total > s.total) as rank
from tt as s;

Answer 2

Something like this maybe:
with tt(S_id,total) as (
select S_id, sum(ca1) + sum(ca2) + sum(exam)
from t
group by S_id
union
select null, 0
)
select s.S_id,
s.total,
(select count(*)+1
from tt as r
where r.total > s.total) as rank
from tt as s
where S_id is not null;

Answer 3

Per my standard Rank Rows answer, use a self join:
with tt (S_id, total) as
(select S_id, sum(ca1 + ca2 + exam) as total
from t group by S_id)
select S.S_id, S.total, 1+count(lesser.total) as RANK
from tt as S
left join tt as lesser
on S.total < lesser.total
group by S.S_id, S.total
order by S.total desc;
S_id total RANK
---------- ---------- ----------
2 198 1
4 198 1
5 183 3
3 165 4
1 143 5
You don't need a CTE; you could use a subquery instead, but you'd have to repeat it.
Using a SELECT clause in a SELECT clause to produce a column (as suggested elsewhere) is AFAIK nonstandard. A self-join is standard and should be easier for the query planner to optimize (if only for that reason). Also, the above query doesn't munge the data: it doesn't add a row to the CTE only to remove it in the main query.
I prefer the sum(ca1 + ca2 + exam) construction to adding the sums. That's how the question was posed, and it asks the system to do less work (only one summation). Sure, addition is commutative, but I wouldn't depend on the query optimizer to notice.