MYSQL math in a single column in a table. Subquery not grouping correctly

Question

I am trying to perform math on a single column in a single table called "actions". The table looks something like this:

Shopper ID   ItemID  Action
1            200     Purchase
1            200     Purchase
1            200     Return
2            100     Purchase
2            100     Return
3            200     Purchase
3            100     Purchase

I want to get a net purchases quantity for each itemID from the above table structure, ie purchases minus returns. Each row is one transaction, so I am using count to derive the quantities. The desired result would look like this:

ItemID   Qty_Purch   Qty_returned  Net_qty_purchased
100      2           1             1
200      3           1             2

My query below has a grouping problem, it adds all the purchases and all the returns in the entire table for each itemID, so the result I get is this:

ItemID   Qty_Purch   Qty_returned  Net_qty_purchased
100      5           2             3
200      5           2             3

My current query looks like this:

create table MYDB.purch0 as
SELECT ItemId,
(Select Count(w1.ItemID) FROM MYDB.actions w1 where w1.action in ("Purchase")) as Qty_Purch,
(Select Count(w2.ItemID) FROM MYDB.actions w2 where w2.action in ("Return  ") ) as Qty_Returned,
(Select  Count(w3.ItemID) FROM MYDB.actions w3 where w3.action in ("Purchase")) -
(Select  Count(w4.ItemID) FROM MYDB.actions w4 where w4.action in ("Return  "))  as Net_qty_purchased

FROM MYDB.actions w
group by itemID

Please help me to get to the desired output. Thanks in anticipation.

Answer 1

You can use conditional aggregation to get the results you want, taking advantage of the fact that MySQL treats a boolean as 1 or 0 in a numeric context:

SELECT ItemID,
       SUM(Action = 'Purchase') AS Qty_purch,
       SUM(Action = 'Return') AS Qty_returned,
       SUM(Action = 'Purchase') - SUM(Action = 'Return') AS Net_qty_purchased
FROM actions
GROUP BY ItemID

Output:

ItemID  Qty_purch   Qty_returned    Net_qty_purchased
100     2           1               1
200     3           1               2

Demo on dbfiddle