根据另一个表中值的存在排除聚合函数中的记录

Question

I'm writing a query that generates statistics based on postcodes and I need to be able to count the number of matching records that are within a range of postcodes except when they exist in a secondary table. This is part of a larger query and I need the count of records for each postcodes in columnar format rather than as separate rows and this minimal example demonstrates what I've attempted:

CREATE TABLE #People
(
    Name nvarchar(10),
    Postcode int
)

INSERT INTO #People VALUES ('Adam', 2000)
INSERT INTO #People VALUES ('John', 2001)
INSERT INTO #People VALUES ('Paul', 2001)
INSERT INTO #People VALUES ('Peter', 2099)
INSERT INTO #People VALUES ('Tom', 4000)

CREATE TABLE #PostcodesToIgnore
(
    Postcode int
)

INSERT INTO #PostcodesToIgnore VALUES (2099)

SELECT SUM(CASE WHEN PostCode BETWEEN 2000 AND 2099 THEN 1 ELSE 0 END) FROM #People

SELECT SUM(CASE WHEN PostCode BETWEEN 2000 AND 2099 
    AND PostCode NOT IN (SELECT PostCode FROM #PostcodesToIgnore) THEN 1 ELSE 0 END) 
    FROM #People

The first query that counts all postcodes within the range works but the second one fails with the error:

Cannot perform an aggregate function on an expression containing an aggregate or a subquery.

While I could refactor the query to include all the criteria from the outer select into each subselect there are quite a few criteria in the real query so I was hoping there might be a more elegant way to go about it?

Answer 1

You could use a left join instead.

SELECT 
SUM
(
    CASE WHEN PostCode BETWEEN 2000 AND 2099 
        AND pcti.PostCode is null 
    THEN 1 
    ELSE 0 
    END
)
FROM #People p
left join #PostcodesToIgnore pcti on pcti.PostCode = p.PostCode

Answer 2

You could remove the SUM and push the query into a derived table or CTE.

The following works

SELECT SUM(PostCodeFlag)
FROM   (SELECT CASE
                 WHEN PostCode BETWEEN 2000 AND 2099
                      AND PostCode NOT IN (SELECT PostCode
                                           FROM   #PostcodesToIgnore) THEN 1
                 ELSE 0
               END AS PostCodeFlag
        FROM   #People) T 

Answer 3

Something like this: Use a CTE to pre-prepare your data, then do a simple grouped count. Or you could have a look on OVER ( https://msdn.microsoft.com/en-us/library/ms189461.aspx )

WITH myCTE AS
(
    SELECT Name,Postcode FROM #People
    WHERE Postcode NOT IN (SELECT Postcode FROM #PostcodesToIgnore)
)
SELECT Postcode, Count(Name) 
FROM myCTE
GROUP BY Postcode       

Answer 4

FROM #people WHERE postcode not in (...).

In fact, it looks like you just don't need any CASE at all and you can specify all of your predicates in the FROM.

Or am I missing something ?