Using aggregate function withouhg using GROUP BY in SQL Server - sql

Question

I have a script that gathers data from a lot of different tables and pull data as I want. This script is long and very sensitive, if I group by anything we might miss on any data being pulled. Is there a way we can use these functions and not have to Group every single value?
Here is the aggregate functions I am trying to use:
CONVERT (INT, ROUND (AVG (CONVERT ( DECIMAL, score)), 0))
This part also uses where clause, in simpler script I usually just have a separate select statement to grab this data but in this case it ties into a lot of other LEFT JOINS so I cant put a Where clause as well.
Here is how I am grabbing this field in single script:
SELECT
CONVERT (INT, ROUND (AVG (CONVERT (DECIMAL, score)), 0)) AS AverageScore
FROM
tbIDs scm
LEFT JOIN
tbIds2 m ON m.ID = scm.ID
WHERE
(Score <> 0) AND (m.Complete= 0)
How can I have this whole statement in another SELECT query?
For example here is how I want to grab this data within another query
SELECT
Firstname, LastName,
CONVERT (INT, ROUND (AVG (CONVERT (DECIMAL, score)), 0)) AS AverageScore
FROM
tbppl P
LEFT JOIN
tbIds ID1 ON P.PPLID = ID1.PPlID
LEFT JOIN
tbIDs2 ID2 ON ID1.ID = ID2.ID
WHERE
(Score <> 0) AND (m.Complete= 0)
When I run it I get an error
is invalid in the select list because it is not contained in either an or the GROUP BY clause.
How can I do this?

Answer 1

Your query should look like:
SELECT
Firstname, LastName,
CONVERT (INT, ROUND (AVG (CONVERT (DECIMAL, score)) OVER (PARTITION BY Firstname, LastName), 0)) AS AverageScore
FROM
tbppl P
LEFT JOIN
tbIds ID1 ON P.PPLID = ID1.PPlID
LEFT JOIN
tbIDs2 ID2 ON ID1.ID = ID2.ID
WHERE
(Score <> 0) AND (m.Complete= 0)
The difference to your query is the part:
OVER (PARTITION BY Firstname, LastName)
which your AVG invoke lacked.
If you specify more columns when using aggregate functions, you need to use their window alternatives (see over clause in SQL) and in PARTITION BY specify additional columns (or use standard AVG function and add GROUP BY clause).

Answer 2

Use over and partition by instead. It applies the aggregate function while preserving the row structure. The format is as follows:
avg(x) over (partition by Group1, Group2, Group3)
Given that you want the average over the full data without any "groups", you simply remove the partition by part.
Your query (without the data type conversions) is as follows:
SELECT
Firstname, LastName,
AVG (score) over () AS AverageScore
FROM
tbppl P
LEFT JOIN
tbIds ID1 ON P.PPLID = ID1.PPlID
LEFT JOIN
tbIDs2 ID2 ON ID1.ID = ID2.ID
WHERE
(Score <> 0) AND (m.Complete= 0)
Edit in response to comment:
If you want to take the average over a subset of rows subject to a certain condition you need to use the following:
avg(case when <conditional logic> then x else null end)
If you only want the output rows populated where the condition is met then use:
case when <conditional logic> then avg(x) end
Combining all of the above for your case gives us:
case when (Score <> 0) AND (m.Complete= 0) then avg(case when when (Score <> 0) AND (m.Complete= 0) then Score else null end) over ()