SQL查询：如何识别组差异

Question

col1	col2	col3
group1	subgroup1	54
group1	subgroup1	31
group1	subgroup2	54
group1	subgroup2	55
group2	subgroup3	40
group2	subgroup3	41
group2	subgroup4	40
group2	subgroup4	41

Each group has a number of subgroups and each subgroup has a number of values.

I need to filter out the group that has the same value in col3. For example, group2 has values(40,41)

only return group1 as subgroup1(54,31) and subgroup2(54,55) shares different values(31, 55) in col3. How can I achieve this in SQL?

Desire result:

col1	col2	col3
group1	subgroup1	54
group1	subgroup1	31
group1	subgroup2	54
group1	subgroup2	55

Answer 1

You could use string_agg(). ie:

select col1, col2, string_agg(col3, ',')
from mytable
group by col1, col2;

CREATE TABLE mytable (
  col1 VARCHAR(10),
  col2 VARCHAR(10),
  col3 varchar(10)
);

INSERT INTO mytable
  (col1, col2, col3)
VALUES
  ('group1', 'subgroup1', '54'),
  ('group1', 'subgroup1', '31'),
  ('group1', 'subgroup2', '54'),
  ('group1', 'subgroup2', '55'),
  ('group2', 'subgroup3', '40'),
  ('group2', 'subgroup3', '41'),
  ('group2', 'subgroup4', '40'),
  ('group2', 'subgroup4', '41');

select col1, col2, string_agg(col3, ',')
from mytable
group by col1, col2;

col1	col2	(No column name)
group1	subgroup1	54,31
group1	subgroup2	54,55
group2	subgroup3	40,41
group2	subgroup4	40,41

DBFiddle demo

EDIT: According to your edited and changed requirement:

with dummy as
(select col1, col2, string_agg(col3, ',') col3
from mytable
group by col1, col2
),
  unq as 
  (
select col3
from dummy
group by col3
having count(*) = 1
)
select * from dummy 
where exists (select col3 from unq where dummy.col3 = unq.col3);

DBFiddle demo