How to fix group by logic in subquery? - sql

Question

I have the following 2 example queries plus their result tables (dummy data) below:
SELECT
subs.Region
,subs.Product
,SUM(p.Price) TotalPriceA
FROM dbo.submission_dtl subs
JOIN dbo.price_dtl p ON subs.SubmissionNumber = p.SubmissionNumber
GROUP BY subs.Region, subs.Product
Region
Product
TotalPriceA
USA
cameras
200
USA
phones
300
Canada
cameras
300
Canada
phones
500
SELECT
r.Region
,r.Product
,SUM(rp.Price) TotalPriceB
FROM dbo.report_dtl r
JOIN dbo.report_price rp ON r.SubmissionNumber = rp.SubmissionNumber
GROUP BY r.Region, rp.Product
Region
Product
TotalPriceB
USA
cameras
201
USA
phones
301
Canada
cameras
301
Canada
phones
501
I want to join them so that the result table resembles this:
Region
Product
TotalPriceA
TotalPriceB
USA
cameras
200
201
USA
phones
300
301
Canada
cameras
300
301
Canada
phones
500
501
But when I used this query, I got a result table that resembled this:
SELECT
subs.Region
,subs.Product
,SUM(p.Price) TotalPriceA
,rptotal.TotalPriceB
FROM dbo.submission_dtl subs
JOIN dbo.price_dtl p ON subs.SubmissionNumber = p.SubmissionNumber
JOIN
(
SELECT
r.Product
,SUM(rp.Price) TotalPriceB
FROM dbo.report_dtl r
JOIN dbo.report_price rp ON r.SubmissionNumber = rp.SubmissionNumber
GROUP BY rp.Product
) rptotal on subs.Product = rptotal.Product
GROUP BY subs.Region, subs.Product, rptotal.TotalPriceB
Region
Product
TotalPriceA
TotalPriceB
USA
cameras
200
502
USA
phones
300
802
Canada
cameras
300
502
Canada
phones
500
802
When I group the subquery by region as well, I get even worse results...

Answer 1

You can try to use two subquery before join
SELECT t1.Region,
t1.Product,
t2.TotalPriceA,
t1.TotalPriceB
FROM (
SELECT
r.Region
,r.Product
,SUM(rp.Price) TotalPriceB
FROM dbo.report_dtl r
JOIN dbo.report_price rp ON r.SubmissionNumber = rp.SubmissionNumber
GROUP BY r.Region, rp.Product
) t1 INNER JOIN (
SELECT
subs.Region
,subs.Product
,SUM(p.Price) TotalPriceA
FROM dbo.submission_dtl subs
JOIN dbo.price_dtl p ON subs.SubmissionNumber = p.SubmissionNumber
GROUP BY subs.Region, subs.Product
) t2 ON t1.Region = t2.Region AND t1.Product = t2.Product

Answer 2

Perhaps a group by is not what is required here, at least not for the final result. Have you considered using the pivot clause instead? As DRapp stated, you might need a union to combine the two queries. Your group by is only required to sumarise the total values before hand, but the pivot should take care of that.
In this example, I'm using a table variable to consolidate all the information and then the pivot. Take a closer look and you'll realise that one of the columns is having a constant all the time for each query. Also, from experience I know that table variables work better with null columns, regardless of the actual data source.
Declare #myData Table (
region varchar(max) null,
product varchar(max) null,
type varchar(max) null,
totalPriceA money null
)
--The type is the constant to know whether it's A or B
Insert Into #myData(region, product, type, totalPrice)
Select Region, Product, 'TotalPriceA', Sum(Price)
From <your tables here>
Group By region, product
--Repeat for total B.
Insert Into #myData(region, product, type, totalPrice)
Select Region, Product, 'TotalPriceB', Sum(Price)
From <your tables here>
Group By region, product
--Now myData table has all the information.
--You only need the output format
Select region, product, TotalPriceA, TotalPriceB
From #myData
Pivot (
Sum(totalPrice)
For type In ('TotalPriceA', 'TotalPriceB')
) As Result
Hope this helps. As you can see, the constant values in column type become the column titles in the final result. You will get null values if one "cell" in the final table doesn't have a corresponding value for that row/column match.