Comparing values from rows with the same group id? - mysql

Question

I'm trying to write a SQL query to do the following:
Given the following table:
+----+----------+-----------+
| id | group_id | value |
+----+----------+-----------+
| 1 | 1 | 0 |
+----+----------+-----------+
| 2 | 1 | 0 |
+----+----------+-----------+
| 3 | 2 | null |
+----+----------+-----------+
| 4 | 3 | -1 |
+----+----------+-----------+
| 5 | 3 | 1 |
+----+----------+-----------+
| 6 | 4 | something |
+----+----------+-----------+
| 7 | 5 | something |
+----+----------+-----------+
select *
where values do not equal each other
group by group_id
For this example, output should be:
+----+----------+-----------+
| id | group_id | value |
+----+----------+-----------+
| 4 | 1 | -1 |
+----+----------+-----------+
| 5 | 1 | 1 |
+----+----------+-----------+
Does anyone know if this is possible?

Answer 1

If you only want to find the group_id values which have different value, you can use this query:
SELECT group_id
FROM data
GROUP BY group_id
HAVING MIN(value) != MAX(value)
Output for your sample data:
group_id
3
If you want to get the rows that are associated with that group_id, use the above query as a subquery for an IN expression:
SELECT *
FROM data
WHERE group_id IN (
SELECT group_id
FROM data
GROUP BY group_id
HAVING MIN(value) != MAX(value)
)
Output
id group_id value
4 3 -1
5 3 1
Demo on dbfiddle

Answer 2

You can do with exists to get group_id which does not have same value. here is the demo.
select
distinct group_id
from data d1
where exists
(
select
group_id
from data d2
where d1.group_id = d2.group_id
and d1.value <> d2.value
)
Output
*--------*
|group_id|
*--------*
| 3 |
*--------*
If you want group_id and value both then try the follwoing
select
group_id,
value
from data d1
where exists
(
select
group_id
from data d2
where d1.group_id = d2.group_id
and d1.value <> d2.value
)
Output:
*------------------*
|group_id | value |
*------------------*
| 3 | -1 |
| 3 | 1 |
*------------------*

Answer 3

This query:
select group_id
from tablename
group by group_id
having count(distinct value) > 1 and count(distinct value) = count(value)
returns the group_ids that you want, so use it with the operator IN:
select * from tablename
where group_id in (
select group_id
from tablename
group by group_id
having count(distinct value) > 1 and count(distinct value) = count(value)
)
See the demo.
Results:
| id | group_id | value |
| --- | -------- | ----- |
| 4 | 3 | -1 |
| 5 | 3 | 1 |