Create 3rd Table SQL with difference of 2 other tables - mysql

Question

Afternoon all
I am trying to create a view with 3 tables.
Create view breakdown as
SELECT x.month_
, x.returns_
, y.sales_
, z.profit_
FROM
(SELECT SUM(Return_Value_Eur) AS returns_
, monthname(Return_Date) AS month_
FROM returns
GROUP BY month_) x
INNER JOIN
(SELECT SUM(Total_Price_Eur) AS sales_
, monthname(Order_Date) AS month_
FROM orders
GROUP BY month_) y ON x.month_ = y.month_
INNER JOIN
(SELECT (SUM(Total_Price_Eur)-SUM(Return_Value_Eur)) AS profit_
, monthname(Order_Date) AS month_
FROM returns, orders
GROUP BY month_) z ON y.month_ = z.month_
I have the following code however I'm struggling with the profit table. This table should be a difference between the y.sales and x.returns. However the figures are coming up as follows, the profit column should be showing 394.
Month_ Returns_ Sales_ Profit_
January 108 502 -251
Any help would be greatly appreciated.

Answer 1

Don't mix the old-school comma syntax for the join operation with the JOIN keyword; actually, just ditch the old-school comma syntax altogether for new queries.
Looks like the problem is the cartesian product of the join between returns and orders, the inline view returning "profit".
It's not clear what we are trying to achieve (broken SQL is like that, in that it fails to accurately communicate the actual specification)
It looks to me (and this is just a guess here) is that the intent is to subtract total returns from total sales, by month.
I'd do it like this:
SELECT v.month_
, IFNULL( SUM(v.returns_) ,0) AS returns_
, IFNULL( SUM(v.sales_ ) ,0) AS sales
, IFNULL( SUM(v.sales_ ) ,0)
- IFNULL( SUM(v.returns_) ,0) AS net_
FROM (
SELECT 0.0 AS sales_
, SUM(r.return_value_eur) AS returns_
, MONTHNAME(r.return_date) AS month_
, MONTH(r.return_date) AS month_num
FROM returns r
GROUP
BY MONTHNAME(r.return_date)
, MONTH(r.return_date)
UNION ALL
SELECT SUM(o.total_price_eur) AS sales_
, 0.0 AS returns_
, MONTHNAME(o.order_date) AS month_
, MONTH(o.order_date) AS month_num
FROM orders o
GROUP
BY MONTHNAME(o.order_date)
, MONTH(o.order_date)
) v
GROUP
BY v.month_
, v.month_num
ORDER
BY v.month_num
Note that this query ignores the year, which is a bit odd, mashing together November 2019, with November 2018, 2017, et al.
The example query orders rows by calendar order, January, February, March, ... that's the reason for the additional month_num column.
If there are no rows in returns or orders for a given month, then that month will not appear in the resultset.
To guarantee rows returned for all 12 calendar months, I would use row source that was guaranteed to return me one row for each month. Then do outer joins to aggregated returns by month and aggregated sales by month. (I'd also have the queries work with the integer value for the month, and then translate the integer to a string "month name" as a final output step.)
Something like this:
SELECT MONTHNAME(STR_TO_DATE(c.month_num,'%m')) AS month_
, IFNULL( tr.returns_ ) ,0) AS returns_
, IFNULL( ts.sales_ ) ,0) AS sales_
, IFNULL( ts.sales_ ) ,0)
- IFNULL( tr.returns_ ) ,0) AS net_
FROM ( -- row source for calendar month 1..12
SELECT 1 AS month_num UNION ALL SELECT 2 UNION ALL SELECT 3 UNION ALL SELECT 4
UNION ALL SELECT 5 UNION ALL SELECT 6 UNION ALL SELECT 7 UNION ALL SELECT 8
UNION ALL SELECT 9 UNION ALL SELECT 10 UNION ALL SELECT 11 UNION ALL SELECT 12
) c
LEFT
JOIN ( -- total returns by month
SELECT SUM(r.return_value_eur) AS returns_
, MONTH(r.return_date) AS month_num
FROM returns r
GROUP
BY MONTH(r.return_date)
) tr
ON tr.month_num = c.month_num
LEFT
JOIN ( -- total orders by month
SELECT SUM(o.total_price_eur) AS sales_
, MONTH(o.order_date) AS month_num
FROM orders o
GROUP
BY MONTH(o.order_date)
) ts
ON ts.month_num = c.month_num
ORDER
BY c.month_num