SQL query: display different group value in different column - sql

Question

I am trying to group by ticker, date and display their value in different columns for different tickers.
But I don't know the exact ticker name which are in my table.
original table(tickers may have other symbols besides A, B, C):
date | ticker | value
-------------------------
1 | A | 5
1 | B | 3
1 | C | 2
2 | A | 5
2 | B | 3
2 | C | 2
3 | A | 5
3 | B | 3
3 | C | 2
.......
Write the SQL query to get the result dataframe:
date | A | B | C
--------------------------------
1 | 5 | 3 | 2
2 | 5 | 3 | 2
3 | 5 | 3 | 2

Answer 1

As sgeddes states, this is accomplished with a pivot. You can create a pivot dynamically when you don't know all the values. I gave an example of doing just that here
https://dba.stackexchange.com/questions/98776/dynamic-select-and-place-result-in-variable-columns/98809#98809
create Table Questions
(
id int identity,
question_id int,
question_name varchar(255)
)
go
with CTEquestion
as
(
select 1 QID
union all
Select QID+1
from CTEquestion
where QID < 11
)
insert questions
select QID, 'Question'+cast(QID as varchar(50))
from CTEquestion
insert QuestionAnswers
values ('2015-04-23', 'a1', 1, 'Canswer1')
, ('2015-04-23', 'a1', 2, 'Ianswer2')
, ('2015-04-23', 'a1', 3, 'Canswer3')
, ('2015-04-23', 'a1', 4, 'Canswer4')
, ('2015-04-23', 'a1', 5, 'Ianswer5')
, ('2015-04-23', 'a1', 6, 'Ianswer6')
, ('2015-04-23', 'a1', 7, 'Canswer7')
, ('2015-04-23', 'a1', 8, 'Canswer8')
, ('2015-04-23', 'a1', 9, 'Canswer9')
, ('2015-04-23', 'a1', 10,'Canswer10')
insert QuestionAnswers
values (CONVERT(DATE, GETDATE()), 'b2', 1, 'Canswer1')
, (CONVERT(DATE, GETDATE()), 'b2', 2, 'Canswer2')
, (CONVERT(DATE, GETDATE()), 'b2', 3, 'Canswer3')
, (CONVERT(DATE, GETDATE()), 'b2', 4, 'Canswer4')
, (CONVERT(DATE, GETDATE()), 'b2', 5, 'Canswer5')
, (CONVERT(DATE, GETDATE()), 'b2', 6, 'Ianswer6')
, (CONVERT(DATE, GETDATE()), 'b2', 7, 'Canswer7')
, (CONVERT(DATE, GETDATE()), 'b2', 8, 'Canswer8')
, (CONVERT(DATE, GETDATE()), 'b2', 9, 'Canswer9')
, (CONVERT(DATE, GETDATE()), 'b2', 10, 'Ianswer10')
insert QuestionAnswers
values (CONVERT(DATE, GETDATE()), 'c3', 1, 'Ianswer1')
, (CONVERT(DATE, GETDATE()), 'c3', 2, 'Ianswer2')
, (CONVERT(DATE, GETDATE()), 'c3', 3, 'Canswer3')
, (CONVERT(DATE, GETDATE()), 'c3', 4, 'Ianswer4')
, (CONVERT(DATE, GETDATE()), 'c3', 5, 'Canswer5')
, (CONVERT(DATE, GETDATE()), 'c3', 6, 'Ianswer6')
, (CONVERT(DATE, GETDATE()), 'c3', 7, 'Canswer7')
, (CONVERT(DATE, GETDATE()), 'c3', 8, 'Canswer8')
, (CONVERT(DATE, GETDATE()), 'c3', 9, 'Canswer9')
, (CONVERT(DATE, GETDATE()), 'c3', 10, 'Ianswer10')
insert QuestionAnswers
values (CONVERT(DATE, GETDATE()), 'a1', 1, 'Canswer1')
, (CONVERT(DATE, GETDATE()), 'a1', 2, 'Ianswer2')
, (CONVERT(DATE, GETDATE()), 'a1', 3, 'Canswer3')
, (CONVERT(DATE, GETDATE()), 'a1', 4, 'Canswer4')
, (CONVERT(DATE, GETDATE()), 'a1', 5, 'Canswer5')
, (CONVERT(DATE, GETDATE()), 'a1', 6, 'Canswer6')
, (CONVERT(DATE, GETDATE()), 'a1', 7, 'Canswer7')
, (CONVERT(DATE, GETDATE()), 'a1', 8, 'Canswer8')
, (CONVERT(DATE, GETDATE()), 'a1', 9, 'Canswer9')
, (CONVERT(DATE, GETDATE()), 'a1', 10, 'Ianswer10')
-->End test data creation
--straight join
select qa.user_id, qa.question_set, q.question_id, qa.answer
from Questions q
join QuestionAnswers qa on qa.question_id=q.question_id
order by qa.user_id
--dynamic pivot
DECLARE
#questionList varchar(max)
, #maxQID int
, #qid int
select #questionList='',#maxQID = MAX(question_id), #qid= MIN(question_id)
FROM Questions
while #qid <= #maxQID
begin
set #questionList=#questionList+'['+cast(#qid as varchar(10))+']'
select #qid=min(question_id)
from Questions
where question_id > #qid
if #qid<=#maxQID
set #questionList=#questionList+', '
end
DECLARE #SQL NVARCHAR(MAX)
SET #SQL = N'
select user_id, '+#questionList+'
from
(select q.question_id, qa.question_set, qa.user_id, qa.answer
from Questions q
join QuestionAnswers qa on qa.question_id=q.question_id) x
PIVOT
(
max(answer)
FOR question_id in ('+#questionList+')
) pvt
order by user_id'
exec sp_executesql #SQL

Answer 2

If you know the ticker values you can do a GROUP BY. Use CASE to do conditional SUM:
select date,
SUM(case when ticker = 'A' then value end) as A,
SUM(case when ticker = 'B' then value end) as B,
SUM(case when ticker = 'C' then value end) as C,
...
from tablename
group by date
Core ANSI SQL-99.
If the different ticker values are unknown, and may change between executions, you need product specific functionality. (The general behavior is that a SELECT always return the same number of columns - independent of table data!)