Display unique combinations with sums for all of one column - sql

Question

EDIT: Sorry for the vagueness here. Actual question: Is there a better more efficient way to get SQL to get the sums for me?
I might be making this harder than I need to, but here is the situation:
I am creating a demographics listing for the heads of the departments where I work. They want to see the typical demographic information. (Age, Gender, Ethnicity, Service member, First time with us, Major, Degree Level etc.)
However when I was doing this I realized that the US census board can change ethnicity on us at any point, so the report would have to be updated at that time, also things like Major and Degree Level could be added or removed based on the Dept of Education.
So I was looking for a way to write a way to display data from a Table but do sums based on that the number of people with the same area in that table. Sounds weird, and I could be making this out to be way worse than I need to. (I am at 1948 lines of code and not even close to being done so far, all of this lines are basically copy/ paste with slight variations)
My statement for pulling the data is perfect. It works and displays exactly how I want it to for the year ranges I want. (Only posting the first year, not the second or fifth as it is just the date range that changes) NOTE: THis is not the complete SELECT state, as the complete isn't needed for this. The column listings are needed, but you all don't really care about where the data comes from or how I am filtering it
DECLARE #StartDate DATETIME
DECLARE #EndDate DATETIME
DECLARE #StartDateLast DATETIME
DECLARE #EndDateLast DATETIME
DECLARE #StartDateFive DATETIME
DECLARE #EndDateFive DATETIME
DECLARE #State VARCHAR(2)
/*Current Year*/
SET #StartDate = '01/01/2013'
SET #EndDate = '12/31/2013'
CREATE TABLE #Demographics /*Table where the data will be processed andultimately displayed from */
(
demo_id INT IDENTITY(1,1) NOT NULL,
header VARCHAR(100),
subheader VARCHAR(100),
value_year1 NUMERIC,
percent_year1 DECIMAL(10,2),
value_year2 NUMERIC,
percent_year2 DECIMAL(10,2),
value_year5 NUMERIC,
percent_year5 DECIMAL(10,2)
)
SELECT s.ID
,g.Name AS [gender]
, FLOOR(DATEDIFF(DAY,s.dob,GETDATE())/365.25) AS Age
, st.Abbreviation
, eg.Name
,f_ss.Name AS [StudentStatus]
,CASE
WHEN EnrollmentStatus.EnrollmentLevel = 'Full Time' THEN 'FT'
WHEN EnrollmentStatus.EnrollmentLevel = 'Part Time' THEN 'PT'
ELSE 'Unknown' END [Enrollment]
,FirstTime.result AS FirstTime
,major.[Abbreviation] AS [Abbreviation]
,major.Major
,mb.Name AS [ServiceBranch]
,FirstDegree.FirstDegree
INTO #DemoTemp1
FROM dbo.Student s
INNER JOIN studentdegree sd ON s.id = sd.StudentID
JOIN dbo.Gender g ON g.id = s.GenderID
INNER JOIN dbo.address ad ON s.AddressID = ad.ID
INNER JOIN dbo.state st ON ad.StateID = st.ID
LEFT JOIN dbo.EthnicGroup eg ON eg.ID = s.EthnicGroupID
LEFT JOIN dbo.MilitaryBranch mb ON mb.ID = s.MilitaryBranchID
JOIN Prospect p ON s.ID = p.StudentID
So I grab the data, and then I dump it into 3 temp tables, and then pull from them to display and store off from another temp table that is defined above as #Demographics.
Here is how the Result-set data should ultimately be displayed:
demo_id| header | subheader | value_year1| percent_year1
1 Gender:| Male | 15195 | 62.00
2 | Female | 9150 | 37.00
3 | Not Disclosed | 23 | 0.00
4 Age: | Under 18 | 2 | 0.00
5 | 18-20 | 142 | 0.00
6 | 21-25 | 1757 | 7.00
7 | 26-30 | 3815 | 15.00
Here is the Insert from the temp tables into the #Demographics table looks right now, and it is the area that I think ultimately should be changed to so that I am not writing out 1050 combinations of degree's, Majors, ages, and what not.
Age and Gender are easy, its the Combination of Degree level and Major that's killing me. There are roughly 750 combinations I need to account for in total.
When I get to things like the Degree/ Majors, it looks like this combo:
INSERT INTO #Demographics
SELECT
''
,'Graduate Certificate in Project Management'
,(SELECT SUM(CASE WHEN dt1.Abbreviation + ' ' + dt1.major ='Certificate Graduate Certificate in Project Management' THEN 1 ELSE 0 END) FROM #DemoTemp1 dt1)
,(SELECT ((SUM(CASE WHEN dt1.Abbreviation + ' ' + dt1.major ='Certificate Graduate Certificate in Project Management THEN 1 ELSE 0 END)*100) / COUNT(dt1.ID)) FROM #DemoTemp1 dt1)
FROM #DemoTemp1 dt1 WHERE dt1.Abbreviation + ' ' + dt1.major ='Certificate Graduate Certificate in Project Management GROUP BY (CASE WHEN dt1.Abbreviation + ' ' + dt1.major ='Certificate Graduate Certificate in Project Management THEN 1 ELSE 0 END)
I have been taking this and editting it multiple times for every Degree level (dt1.abbreviation) and Major (dt1.major)
How do I basically take the dt1.Abbreviation + ' ' + dt1.major combination and get SQL to spit out the sums of the students in that without coding for every combination myself. With the result set I posted above as the desired format
EDIT: This is what I am using now. I am just tweaking the PERC/ Y2P/ Y5P columns to show an actual percentage and not a decimal
SELECT DISTINCT(dt.Abbreviation + ' ' + m.Name) [Program]
,YearOne.Counts
,YearOne.Perc
,YearTwo.Counts [y2c]
,YearTwo.Perc [y2p]
,YearFive.Counts [y5c]
,YearFive.perc [y5p]
FROM degree d
INNER JOIN degreetype dt ON d.DegreeTypeID = dt.ID
INNER JOIN dbo.Majors m ON d.MajorID = m.id
OUTER APPLY (SELECT count(*) Counts
,CAST((CAST(COUNT(*)AS float) * 100) / CAST((SELECT count(dt2.id)FROM #DemoTemp1 dt2)AS float)as decimal(10,2))*100 Perc
--,(SELECT count(dt2.id) [dude]FROM #DemoTemp1 dt2) perc2
FROM #DemoTemp1 dt1
WHERE dt1.[DegreeType]+' '+ dt1.Major = dt.name +' '+ m.name
) YearOne
OUTER APPLY (SELECT count(*) Counts
,CAST((CAST(COUNT(*)AS float) * 100) / CAST((SELECT count(dt3.id)FROM #DemoTemp2 dt3)AS float)as decimal(10,2))*100 Perc
--,(SELECT count(dt2.id) [dude]FROM #DemoTemp1 dt2) perc2
FROM #DemoTemp2 dt2
WHERE dt2.[DegreeType]+' '+ dt2.Major = dt.name +' '+ m.name
) YearTwo
OUTER APPLY (SELECT count(*) Counts
,CAST((CAST(COUNT(*)AS float) * 100) / CAST((SELECT count(dt4.id)FROM #DemoTemp5 dt4)AS float)as decimal(10,2))*100 Perc
--,(SELECT count(dt2.id) [dude]FROM #DemoTemp1 dt2) perc2
FROM #DemoTemp5 dt5
WHERE dt5.[DegreeType]+' '+ dt5.Major = dt.name +' '+ m.name
) YearFive
WHERE 1=1
AND d.Archived = 0
GROUP BY (dt.Abbreviation + ' ' + m.Name), YearOne.Counts, YearOne.Perc,YearTwo.Counts,YearTwo.Perc,YearFive.Counts ,YearFive.perc
HAVING YearOne.Counts > 0

Answer 1

You could change your first query to something like this. It is the same thing and a lot simpler.
SELECT
'Gender'
,'Male'
, count(*)
, (COUNT(*) * 100) / COUNT(dt1.ID)
FROM #DemoTemp1 dt1
WHERE dt1.gender = 'Male'
GROUP BY dt1.gender
You could do something similar in the second one. No need to hit the same table over and over to get aggregate data.
OK. After all the back and forth I think I finally understand what you are asking.
Does something like this work for you second query?
SELECT
''
, dt1.Abbreviation + ' ' + dt1.major
, count(*)
, (count(*) * 100) / COUNT(dt1.ID))
FROM #DemoTemp1 dt1
GROUP BY dt1.Abbreviation + ' ' + dt1.major