Reputation: 47280
What is the best way to to perform this query. I have the following table
mytable with columns
x y
1 a
2 b
3 c
and I would like to (in pseudo sql)
select x as x1 ,x as x2, x as x3 from mytable where ????
when
x1 is x where y=a
x2 is x where y=b
x3 is x where y=c
so I would like as a result
1, 2, 3
I am currently using cte's and and a very large dataset, I am trying to reduce the query time, is it always necessary to have 3 table scans ?
Upvotes: 8
Views: 47295
Reputation: 5050
MySQL 5.5.32 Schema Setup:
CREATE TABLE Table1
(`x` int, `y` varchar(1))
;
INSERT INTO Table1
(`x`, `y`)
VALUES
(1, 'a'),
(2, 'b'),
(3, 'c')
;
CREATE TABLE mytable
(`x` int, `y` varchar(1))
;
INSERT INTO mytable
(`x`, `y`)
VALUES
(1, 'a'),
(2, 'b'),
(3, 'c')
;
Query 1:
SELECT x1.x as x1, x2.x as x2, x3.x as x3
FROM mytable x1
INNER JOIN mytable x2 ON x2.y='b'
INNER JOIN mytable x3 ON x3.y='c'
WHERE x1.y='a'
| X1 | X2 | X3 |
|----|----|----|
| 1 | 2 | 3 |
Upvotes: 0
Reputation: 3034
I would of done this :
SELECT
tableRowA.x as x1
tableRowB.x as x2
tableRowC.x as x3
FROM
table as tableRowA,
table as tableRowB,
table as tableRowC
WHERE
tableRowA.y = 1
tableRowB.y = 2
tableRowC.y = 3
Bit easier to understand, and pull out more info if you need multiple columns from each row
Upvotes: 5
Reputation: 76
From your question it appears you would like the last three instances of a condition, or tie three different conditions together. Would the following example satisfy your question:
mytable:
(unique keys 1..n) (col1)
student-id | course-id | grade
s1 gen101 g1
s1 cmp202 g2
s1 psy303 g3
s1 c4 g4
s2 c1 g5
Lets say we only want the students which have three specific courses (gen101, cmp202, and psy303) and show those grades ignoring anyone else.
select gen.student-id as student-id
, gen.grade as gen101-gr
, cmp.grade as cmp202-gr
, psy.grade as psy303-gr
from mytable gen
, mytable cmp
, mytable psy
where gen.course-id = 'gen101'
and gen.student-id = cmp.student-id
and cmp.course-id = 'cmp202'
and cmp.studnet-id = psy.student-id
and psy.course-id = 'psy303'
This should give one row:
student-id gen101-gr cmp202-gr psy303-gr
s1 g1 g2 g3
Hope that gives you enough to work on.
Upvotes: 1
Reputation:
In the example given, there are only 3 rows of input and one row of output. I assume that there is going to be at least one other column involved, such that input data:
w x y
---------
w1 1 a
w1 2 b
w1 3 c
w2 4 a
w2 5 b
w2 6 c
.
.
.
is to become output:
w x1 x2 x3
-----------
w1 1 2 3
w2 4 5 6
.
.
.
This can be done in a single pass using a query like:
select w,
max(case when y = 'a' then x end) x1,
max(case when y = 'b' then x end) x2,
max(case when y = 'c' then x end) x3
from datatable
where y in ('a','b','c')
group by w
Upvotes: 3
Reputation: 2293
My recommendation would be to select the results ordered or grouped by column y and the use that information to split the resultset into several lists for an application to process. If you want to do this only in the database I'm afraid multiple table scans (or joins) are necessary.
Another fix is to migrate the information in column y to another table (with a foreign key reference), to be able to join more effectively to that.
Upvotes: 0
Reputation: 45902
Another solution:
SELECT x, y FROM table WHERE y IN ('a', 'b')
You will have a result set:
x | y
-----
1 | a
2 | b
This result set can then be used in an application to get desired result.
Upvotes: 1
Reputation: 45902
You should use 3 queries. It will be a lot faster with proper indexing when self joins. Additionally it will be more readable.
If you would like one query call, it might be this :)
SELECT
(SELECT x FROM table WHERE y=1) AS x1,
(SELECT x FROM table WHERE y=2) AS x2,
(SELECT x FROM table WHERE y=3) AS x3
Upvotes: 15
Reputation: 1594
SELECT Case When y = 1 Then x1 When y = 2 Then x2 Else x3 End FROM mytable
Upvotes: 0