SQL: union multiple single row results sets from the same query - sql-server

Question

I have a query that, when given a specific date, examines a large dataset (around 36m and growing) and returns an aggregate. Everything is working as expected... However, my end goal is to be able to determine a yearly average of these values. Perhaps my brain is running on empty, but I'm trying to do this dynamically where I don't have to run the query 365 times and then average.....
I need to find the yearly average of the 365 results, per #inst, per #program.
Any pointers in the right direction would be most appreciated.
Query:
USE HCODS
GO
DECLARE #user_date DATETIME
DECLARE #inst VARCHAR(4)
DECLARE #program VARCHAR(4)
SELECT #user_date = '9/30/2016'
SELECT #inst = 'SAC'
SELECT #program = 'PSU';
WITH T AS (
SELECT
B.OFFENDERID
,Institution = I.ORGCOMMONID
,BedUse = B.BEDUSE
,BeginEffectiveDtTm = CAST(B.BEDASSIGNMENTDATE AS DATETIME) + CAST(B.BEDASSIGNMENTTIME AS DATETIME)
,EndEffectiveDtTm = CASE WHEN B.BEDASSIGNMENTSTATUS = 'U' THEN
(CAST(B.INMBEDSTATUSDATE AS DATETIME) + CAST(B.INMBEDSTATUSTIME AS DATETIME)) ELSE NULL END
FROM ODS.BEDASSIGNMENT (NOLOCK) B
INNER JOIN (
SELECT F.PARTYID, I.ORGCOMMONID
FROM ODS.ORGANIZATIONPROF (NOLOCK) AS F
INNER JOIN ODS.ORGANIZATIONPROF (NOLOCK) AS I ON F.ORGAREACODE = I.PARTYID
) AS I ON B.FACILITYWHEREBEDLOCATED = I.PARTYID
WHERE B.BEDASSIGNMENTDATE BETWEEN '1/1/2016' AND '12/31/2016'
AND B.BEDASSIGNMENTSTATUS IN ('U','M')
)
SELECT CAST(#user_date AS DATE)
,T.INSTITUTION
,T.BEDUSE
,COUNT(*)
FROM T
WHERE
(
(
T.BEGINEFFECTIVEDTTM <= DATEADD(second,-1,(#user_date+1))
AND
T.ENDEFFECTIVEDTTM >= #user_date
)
OR T.ENDEFFECTIVEDTTM IS NULL
)
AND T.INSTITUTION = #inst
AND T.BedUse = #program
GROUP BY
T.Institution
,T.BedUse
Result sets (each one obtained by a single running of the query)
Date |Institution |BedUse |Count
-----------|------------|-------|-------
2016-09-30 |SAC |PSU |446
2016-10-01 |SAC |PSU |421
2016-10-02 |SAC |PSU |423
etc......

Answer 1

While it is hard to answer the question of your data without seeing it. I can turn you onto a SQL concept of windowed functions. This in essense is doing an inline grouping to aggregate data. If I have a single set but want to fashion multiple statements over it to see different things, this statement is perfect.
So in an example I am in essence going from the 1st of January of 2015 to today(dynamic as this could be any day even after I post this). I am then picking a random number of 1 to 100 to populate my row of data in my temporary set with. I then can do my aggregate operations on that.
DECLARE #Data TABLE ( Id INT IDENTITY, val INT, dt DATETIME)
DECLARE #Start DATETIME = '1-1-2015'
SET NOCOUNT ON;
WHILE #Start <= GETDATE()
BEGIN
INSERT INTO #Data VALUES (ABS(CHECKSUM(NewId())) % 100, #Start)
SELECT #Start = DATEADD(DAY, 1, #STart)
END
SELECT DISTINCT
SUM(Val) OVER() AS TotalValues
, COUNT(*) OVER() AS rowCounts
, DATEADD(YEAR, DATEDIFF(YEAR, 0, Dt), 0) AS YearDate
, COUNT(*) OVER(PARTITION BY DATEADD(YEAR, DATEDIFF(YEAR, 0, Dt), 0)) AS DaysInYear
, SUM(Val) OVER(PARTITION BY DATEADD(YEAR, DATEDIFF(YEAR, 0, Dt), 0)) AS ValsByYear
, AVG(Val) OVER(PARTITION BY DATEADD(YEAR, DATEDIFF(YEAR, 0, Dt), 0)) AS AVGByYear
, DATEADD(Month, DATEDIFF(Month, 0, Dt), 0) AS MonthDate
, COUNT(*) OVER(PARTITION BY DATEADD(Month, DATEDIFF(Month, 0, Dt), 0)) AS DaysInMonth
, SUM(Val) OVER(PARTITION BY DATEADD(Month, DATEDIFF(Month, 0, Dt), 0)) AS ValsByMonth
, AVG(Val) OVER(PARTITION BY DATEADD(Month, DATEDIFF(MOnth, 0, Dt), 0)) AS AVGByMonth
From #Data