SQL - Counting column categories - sql

Question

I am new to SQL, but I am trying to create a report where I can count the food purchases made by students and categorized by the Race/Ethnicity and choice of food every year.
For Example:
2012 2013
Student Vegetarian Meat Unknown Vegetarian Meat Unknown
---------------------------------------------------------------------------
Black 5 Purchases 4 Purchases 3 Purchases 5 Purchases etc
White 4 Purchases 3 Purchases 3 Purchases etc etc
Asian 4 Purchases 1 Purchases 6 Purchases etc etc
While I was able to get the races in one column, I am lost at getting the count on different food categories. All I know is that the pur_type is classified as "Veg", "Me" and "Ukt" for Vegetarian, Meat and unknown purchase.
Here is what I tried to do:
select
DATEPART(YYYY,pur_date) AS Year
,count (*) Citations
, race
, sex
, ethnicity
, GETDATE() as YTD
,COUNT(CASE WHEN pur_type = 'Veg' THEN 'Vegetarian'
WHEN pur_type = 'Me' THEN 'Meat'
WHEN pur_type = 'UKt' THEN 'Unknown'
ELSE ''
End as "Type"
, CASE
WHEN race='W' and sex='M' and ethnicity <> 'H' THEN 'Caucasian(Male)'
WHEN race='W' and sex='F' and ethnicity <> 'H' THEN 'Caucasian(Female)'
WHEN race='B' and sex='M' and ethnicity <> 'H' THEN 'African-American(Male)'
WHEN race='B' and sex='F' and ethnicity <> 'H' THEN 'African-American(Female)'
--WHEN race='A' and sex='M' and ethnicity <> 'H' THEN 'Pacific Islander(Male)'
--WHEN race='A' and sex='F' and ethnicity <> 'H' THEN 'Pacific Islander(Female)'
WHEN race='A' and sex='M' and ethnicity <> 'H' THEN 'Asian(Male)'
WHEN race='A' and sex='F' and ethnicity <> 'H' THEN 'Asian(Female)'
ELSE ''
END as 'Race'
from purchase
join purcharge
on purchase.purchaseid = purcharge.purchaseid
where pur_date between #startDate and dateadd(DD, +1, GETDATE())
and pur_type in ('Veg', 'Me', 'UKt')
GROUP BY DATEPART(YYYY,pur_date), race, sex, ethnicity , pur_type
ORDER BY 'Year','Race/Sex'
Looking for guidance. I am thinking maybe a sub query would work, maybe more case statements, but I am not able to have a count of how many food categories were purchased by students in each ethnic group.
Also,I am trying to get these results in an SSRS report so it may look slightly different in a SQL query.

Answer 1

You could look into the PIVOT statement but if it was me, I'd do more CASE...WHEN and then GROUP BY.
So, instead of
COUNT(CASE WHEN pur_type = 'Veg' THEN 'Vegetarian' (etc) AS Type
I'd do
SUM(CASE WHEN pur_type = 'Veg' THEN 1 ELSE 0) AS Vegetarian
etc.

Answer 2

This isn't really the best way to create a report, but if you need to use SQL (as opposed to having another layer of code to generate the report from a bunch of subsets of data), you could use subqueries. Without seeing the table structure, it is hard to write the query, but it could be something like this:
Select
race
, (select count(*) from purchase where pur_type = 'Veg' and race = race) as Vegetables
, (select count(*) from purchase where pur_type = 'Me' and race = race) as Meat
, (select count(*) from purchase where pur_type = 'UKt' and race = race) as Unknown
From purchase
Like I said, this isn't ideal, but it would get your the distinct subsets of data.

Answer 3

I don't know what your data looks like and havn't got time to build sample data but I think you've actually just got a typo or two..
SELECT
DATEPART(YYYY,pur_date) AS [Year]
,count (*) Citations
, race
, sex
, ethnicity
, GETDATE() as YTD
, COUNT(*) AS [PurchaseCount]
, CASE
WHEN pur_type = 'Veg' THEN 'Vegetarian'
WHEN pur_type = 'Me' THEN 'Meat'
WHEN pur_type = 'UKt' THEN 'Unknown'
ELSE ''
End as [Type]
, CASE
WHEN race='W' and sex='M' and ethnicity <> 'H' THEN 'Caucasian(Male)'
WHEN race='W' and sex='F' and ethnicity <> 'H' THEN 'Caucasian(Female)'
WHEN race='B' and sex='M' and ethnicity <> 'H' THEN 'African-American(Male)'
WHEN race='B' and sex='F' and ethnicity <> 'H' THEN 'African-American(Female)'
--WHEN race='A' and sex='M' and ethnicity <> 'H' THEN 'Pacific Islander(Male)'
--WHEN race='A' and sex='F' and ethnicity <> 'H' THEN 'Pacific Islander(Female)'
WHEN race='A' and sex='M' and ethnicity <> 'H' THEN 'Asian(Male)'
WHEN race='A' and sex='F' and ethnicity <> 'H' THEN 'Asian(Female)'
ELSE ''
END as [Race]
FROM purchase
JOIN purcharge
ON purchase.purchaseid = purcharge.purchaseid
WHERE pur_date BETWEEN #startDate AND dateadd(DD, +1, GETDATE())
AND pur_type in ('Veg', 'Me', 'UKt')
GROUP BY DATEPART(YYYY,pur_date), race, sex, ethnicity , pur_type
ORDER BY [Year],[Race], [Type]
This will not give exactl what you asked for as it includes both race and gender in your [Race] case statement but maybe that's what you really wanted, I don't know...
Anyway, all I really changed was the count(*) as [PurchaseCount] , the following line's CASE statement and the GROUP BY clause. COUNT(*) will just give you the count of records that fall within your GROUP BY clause. In this case a count of records for every [Year], [Race] and [Type]