Count Unique Identifiers by a result in another column - sql

Question

I am trying to return some values in SQL. Let's call Column A as UniqueIdentifier I then have a row that lists two different actions that can be done to this unique identifier, let's call the column action and the actions a and b
The actions can be duplicated more than one time per unique identifier, thus listing tons of duplicated UniqueIdentifier
How do I get the count of UniqueIdentifier where action a has been performed and the count where action a has not?
I did create a really long and cumbersome way(listed below) using temp tables but I feel like there must be a more roundabout way
<!-- language: lang-sql -->
select
UniqueIdentifier
,case when action = 'a' then 1 else 0 end [actionflag]
into #actionflags
from mydatabase
select
distinct UniqueIdentifier
,sum(actionflag)
into #actionflagscount
from #actionflags
group by UniqueIdentifier
select case when actionflag > 0 then 1 else 0 end [actionflag]
,count(uniqueidentifier)
from #actionflagscount
group by case when actionflag > 0 then 1 else 0 end

Answer 1

Do you want DISTINCT ?
select case when actionflag > 0 then 1 else 0 end [actionflag],
count(distinct uniqueidentifier)
from #actionflagscount
group by case when actionflag > 0 then 1 else 0 end

Answer 2

This is what I came up with
DROP TABLE IF EXISTS #temp
CREATE TABLE #temp ([UniqueIdentifier] INT, Action NVARCHAR(10))
INSERT INTO #temp ([UniqueIdentifier], Action) VALUES (1, 'a')
INSERT INTO #temp ([UniqueIdentifier], Action) VALUES (2, 'a')
INSERT INTO #temp ([UniqueIdentifier], Action) VALUES (3, 'a')
INSERT INTO #temp ([UniqueIdentifier], Action) VALUES (4, 'b')
INSERT INTO #temp ([UniqueIdentifier], Action) VALUES (5, 'b')
INSERT INTO #temp ([UniqueIdentifier], Action) VALUES (6, 'c')
INSERT INTO #temp ([UniqueIdentifier], Action) VALUES (7, 'c')
SELECT CASE WHEN Action = 'a' THEN 1 ELSE 0 END AS ActionFlag, COUNT(*)
FROM #temp
GROUP BY CASE WHEN Action = 'a' THEN 1 ELSE 0 END
-- Results:
-- ActionFlag (No column name)
-- 0 4
-- 1 3

Answer 3

Considering you may have:
The same action performed several times on the same UniqueIdentifier
Different actions performed on the same UniqueIdentifier
I would do:
SELECT DISTINCT UniqueIdentifier, CASE WHEN EXISTS (SELECT 1 FROM MyDatabase WHERE action = 'a' and UniqueIdentifier = MDB.UniqueIdentifier) THEN 1 ELSE 0 END AS ActionPerformed
FROM MyDatabase MDB
If this looks fine, then, a simple COUNT will do the trick surrounding the previous query.
SELECT Count(*), ActionPerformed
FROM (
SELECT DISTINCT UniqueIdentifier, CASE WHEN EXISTS (SELECT 1 FROM MyDatabase WHERE action = 'a' and UniqueIdentifier = MDB.UniqueIdentifier) THEN 1 ELSE 0 END AS ActionPerformed
FROM MyDatabase MDB
) T
GROUP BY ActionPerformed
ActionPerformed = 1 means action a was performed (possibly action b was performed too on the same UniqueIdentifier). ActionPerformed = 0 means action a was not performed (but it does not say anything about action b)

Answer 4

I would use two levels of aggregation:
select sum(case when num_as > 0 then 1 else 0 end) as num_with_as,
sum(case when num_as = 0 then 1 else 0 end) as num_without_as
from (select uniqueidentifier,
sum(case when action = 'a' then 1 else 0 end) as num_as
from mydatabase
group by uniqueidentifier
) d;
The inner query counts the number of "a"s per uniqueidentifier. The outer query uses this to get the two values you want.