Getting the sum of columns based on row values - sql

Question

I have a table that looks like the following.
EMPNUM EMPNAME LOCATION CATEGORY COUNT
123 JOHN DOE BLDG A 1 5
123 JOHN DOE BLDG A 1 6
123 JOHN DOE BLDG A 2 4
123 JOHN DOE BLDG A 3 7
123 JOHN DOE BLDG B 1 1
123 JOHN DOE BLDG B 2 3
234 EMILY DOE BLDG A 1 1
234 EMILY DOE BLDG A 2 2
234 EMILY DOE BLDG A 3 4
234 EMILY DOE BLDG B 2 3
234 EMILY DOE BLDG B 2 9
234 EMILY DOE BLDG B 3 3
I would like to transport it into columns that will yield to an output similar to below. I need to get the sum of COUNT based on the values of LOCATION and CATEGORY
EMPNUM EMPNAME SUM_A1 SUM_A2 SUM_A3 SUM_B1 SUM_B2 SUM_B3
123 JOHN DOE 11 4 7 1 3 0
234 EMILY DOE 1 2 4 0 12 3
Is there any way to do this as an SQL query? or in Crystal reports (though I prefer output using SQL)

Answer 1

If you are using 11g or later try
select * from table1
PIVOT (SUM("COUNT")
FOR ("LOCATION","CATEGORY") IN
(('BLDG A',1) AS sum_a1,
('BLDG A',2) AS sum_a2,
('BLDG A',3) AS sum_a3,
('BLDG B',1) AS sum_b1,
('BLDG B',2) AS sum_b2,
('BLDG B',3) AS sum_b3));
Here is a fiddle
Otherwise use APC's solution

Answer 2

This will work providing the values in LOCATION and CATEGORY are constant:
select empnum
, empname
, sum(case when location='BLDG A' and category = 1 then count else 0 end) sum_a1
, sum(case when location='BLDG A' and category = 2 then count else 0 end) sum_a2
, sum(case when location='BLDG A' and category = 3 then count else 0 end) sum_a3
, sum(case when location='BLDG B' and category = 1 then count else 0 end) sum_b1
, sum(case when location='BLDG B' and category = 2 then count else 0 end) sum_b2
, sum(case when location='BLDG B' and category = 3 then count else 0 end) sum_b3
from your_table
group by empnum
, empname
If the values are not known or not stable when you run the query you will need to use dynamic SQL.
Note that if you are on 11g you should employ A B Cade's PIVOT solution, which is more elegant.

Answer 3

The other answers will work great if you have a known number of values to transform into columns. But if you have an unknown number, then you can use dynamic sql to generate the results.
You would create the following procedure:
CREATE OR REPLACE procedure test_dynamic_pivot(p_cursor in out sys_refcursor)
as
sql_query varchar2(1000) := 'select empnum, empname';
begin
for x in (select distinct location, category from yourtable order by 1)
loop
sql_query := sql_query ||
' , sum(case when location = '''||x.location||''' and category='||x.category||' then cnt else 0 end) as sum_'||substr(x.location, -1, 1)||x.category;
dbms_output.put_line(sql_query);
end loop;
sql_query := sql_query || ' from yourtable group by empnum, empname';
open p_cursor for sql_query;
end;
/
And then to execute it:
variable x refcursor
exec test_dynamic_pivot(:x)
print x
The result is the same as the hard-coded version:
| EMPNUM | EMPNAME | SUM_A1 | SUM_A2 | SUM_A3 | SUM_B1 | SUM_B2 | SUM_B3 |
----------------------------------------------------------------------------
| 234 | EMILY DOE | 1 | 2 | 4 | 0 | 12 | 3 |
| 123 | JOHN DOE | 11 | 4 | 7 | 1 | 3 | 0 |