Oracle pivot - one table field split into rows and columns - sql

Question

I have an Oracle 12c table that looks something like this:
Name Class Type
--------------------------
Alice Math S
Alice Female A
Bob Anthropology S
Bob Male A
Charlie Science S
Charlie Tennis A
Charlie Male A
Do Math S
Do Female A
Do Tennis A
Elmer Male A
Elmer Science S
I would like to pivot the table to produce a report that looks like below. Classes of type "S" are rows, and type "A" are columns. And then count the number of people that fall into each category. For example, there are two people that are both Female and take class Math - Alice and Do.
Female Male Tennis Total
-----------------------------------------------
Math 2 0 1 3
Anthropology 0 1 0 1
Science 0 2 1 3
Total 2 3 2 7
I am familiar with pivot and cube in Oracle but I am stuck with getting the Class field split between rows and columns in the report.
I would really appreciate your help.

Answer 1

You can use conditional aggregation as following:
SQL> with dataa(name, class, type) as
2 (
3 select 'Alice', 'Math' ,'S' from dual union all
4 select 'Alice', 'Female' ,'A' from dual union all
5 select 'Bob', 'Anthropology' ,'S' from dual union all
6 select 'Bob', 'Male' ,'A' from dual union all
7 select 'Charlie', 'Science' ,'S' from dual union all
8 select 'Charlie', 'Tennis' ,'A' from dual union all
9 select 'Charlie', 'Male' ,'A' from dual union all
10 select 'Do', 'Math' ,'S' from dual union all
11 select 'Do', 'Female' ,'A' from dual union all
12 select 'Do', 'Tennis' ,'A' from dual union all
13 select 'Elmer', 'Male' ,'A' from dual union all
14 select 'Elmer', 'Science' ,'S' from dual
15 ),
16 -- your query starts from here
17 CTE AS
18 (
19 SELECT S.CLASS AS CLASS,
20 COALESCE(SUM(CASE WHEN A.CLASS = 'Female' THEN 1 END),0) AS Female,
21 COALESCE(SUM(CASE WHEN A.CLASS = 'Male' THEN 1 END),0) AS Male,
22 COALESCE(SUM(CASE WHEN A.CLASS = 'Tennis' THEN 1 END),0) AS Tennis,
23 COALESCE(SUM(CASE WHEN A.CLASS IN ('Female','Male','Tennis') THEN 1 END),0) AS TOTAL
24 FROM DATAA S JOIN DATAA A ON (A.name = S.name)
25 WHERE S.type = 'S' AND A.TYPE = 'A'
26 GROUP BY S.CLASS
27 )
28 SELECT CLASS, Female, Male, Tennis, TOTAL FROM CTE
29 UNION ALL
30 SELECT 'Total' AS CLASS, SUM(Female), SUM(Male), SUM(Tennis), SUM(TOTAL) FROM CTE;
CLASS FEMALE MALE TENNIS TOTAL
------------ ---------- ---------- ---------- ----------
Science 0 2 1 3
Anthropology 0 1 0 1
Math 2 0 1 3
Total 2 3 2 7
SQL>
Cheers!!

Answer 2

You have the destination "row" value and the destination "column" value in separate rows. In order to do a GROUP BY CUBE, you need to put them together in the same input row using a JOIN. So do a JOIN, then a GROUP BY CUBE, then a PIVOT. To keep things straight, use the GROUPING function to see when you are grouping "rows" and when you are grouping "columns". This becomes a lot of work, and you wind up getting NULLs in some places where you want zeroes. Here is the solution for curiosity's sake:
with data(name, class, type) as (
select 'Alice', 'Math' ,'S' from dual union all
select 'Alice', 'Female' ,'A' from dual union all
select 'Bob', 'Anthropology' ,'S' from dual union all
select 'Bob', 'Male' ,'A' from dual union all
select 'Charlie', 'Science' ,'S' from dual union all
select 'Charlie', 'Tennis' ,'A' from dual union all
select 'Charlie', 'Male' ,'A' from dual union all
select 'Do', 'Math' ,'S' from dual union all
select 'Do', 'Female' ,'A' from dual union all
select 'Do', 'Tennis' ,'A' from dual union all
select 'Elmer', 'Male' ,'A' from dual union all
select 'Elmer', 'Science' ,'S' from dual
)
, joined_data as (
select s.class row_class, a.class col_class
from data s
join data a
on s.type = 'S' and a.type = 'A' and s.name = a.name
)
, cubed_data as (
select case grouping(col_class) when 1 then 'Total' else col_class end col_class,
row_class, grouping(row_class) gr_row, count(*) cnt
from joined_data
group by cube(row_class, col_class)
)
select case gr_row when 1 then 'Total' else row_class end " ",
"Female", "Male", "Tennis", "Total"
from cubed_data
pivot(
sum(cnt) for col_class in (
'Female' as "Female",'Male' as "Male",'Tennis' as "Tennis",'Total' as "Total"
)
)
order by gr_row, row_class;
Female Male Tennis Total
------------ ---------- ---------- ---------- ----------
Anthropology 1 1
Math 2 1 3
Science 2 1 3
Total 2 3 2 7
Regards,
Stew

Answer 3

For this requirement, I prefer the solution by Tejash to the PIVOT demonstration in my other answer. However, I would suggest using ROLLUP instead of UNION ALL. Two lines are changed and the UNION ALL is left out.
with data(name, class, type) as (
select 'Alice', 'Math' ,'S' from dual union all
select 'Alice', 'Female' ,'A' from dual union all
select 'Bob', 'Anthropology' ,'S' from dual union all
select 'Bob', 'Male' ,'A' from dual union all
select 'Charlie', 'Science' ,'S' from dual union all
select 'Charlie', 'Tennis' ,'A' from dual union all
select 'Charlie', 'Male' ,'A' from dual union all
select 'Do', 'Math' ,'S' from dual union all
select 'Do', 'Female' ,'A' from dual union all
select 'Do', 'Tennis' ,'A' from dual union all
select 'Elmer', 'Male' ,'A' from dual union all
select 'Elmer', 'Science' ,'S' from dual
)
SELECT case grouping(S.CLASS) when 1 then 'Total' else s.class end AS CLASS,
COALESCE(SUM(CASE WHEN A.CLASS = 'Female' THEN 1 END),0) AS Female,
COALESCE(SUM(CASE WHEN A.CLASS = 'Male' THEN 1 END),0) AS Male,
COALESCE(SUM(CASE WHEN A.CLASS = 'Tennis' THEN 1 END),0) AS Tennis,
count(*) AS TOTAL
FROM DATA S JOIN DATA A ON (A.name = S.name)
WHERE S.type = 'S' AND A.TYPE = 'A'
GROUP BY rollup(S.CLASS);
CLASS FEMALE MALE TENNIS TOTAL
------------ ---------- ---------- ---------- ----------
Anthropology 0 1 0 1
Math 2 0 1 3
Science 0 2 1 3
Total 2 3 2 7
Regards,
Stew