SQL: 4 Tables to 1 table with counts, groups and deductions?! - mysql

Question

I have a project with lost, found and matched luggage on airports. I made it in Java(FX) and mySQL.
This is what I have:
I have 4 tables:
1 table Airports with 2 columns:
Airport_id & Airport_name
1 table Found with 3 columns:
Found_id & Found_AirportID & Matched
1 table Lost with 3 columns:
Lost_id & Lost_AirportID & Matched
1 table Match with 3 columns:
Match_id & Match_LostID & Match_FoundID & Match_AirportID
Whenever a match is made, the Match table gets a new row with the Match_LostID (from the Lost_id) & Match_FoundID (from the Found_id) and the Match_AirportID (Found_AirportID)
The Matched (in both Found & Lost) get both set to 1, instead of NULL
All the AirportID's are linked to the Airport table.
What I want;
For each and every airport I want the count of the lost items, the count of found items and the count of matched items. BUT when a item is 'matched' it may not appear in the count of the lost and found.
So I want a table with 4 columns:
Airportname, Count of Found, Count of Lost, Count of Matched.
I've made the following Query:
SELECT vv.Airport_name,
COUNT(DISTINCT gb.Found_id) countFound,
COUNT(DISTINCT vb.Lost_id) countLost,
COUNT(DISTINCT kt.Match_id) countMatch
FROM Airports vv
LEFT JOIN Found gb ON vv.Airport_id = gb.Found_AirportID
LEFT JOIN Lost vb ON vv.Airport_id = vb.Lost_AirportID
LEFT JOIN Match kt ON vv.Airport_id = kt.Match_AirportID
WHERE vb.Matched IS NULL OR gb.Matched IS NULL
GROUP BY vv.Airport_name
I manage to get all the count items for Found, Lost and Match.
e.g. New York has 2 found, 2 lost and 1 match.
This is displayed correctly in the table.
But as I said, if there is a match it should be removed from found and lost. It should be:
New York has 1 found, 1 lost and 1 match.
I tried a lot of things, 1 time I manage to do it but then a Airport is missing or it gets deducted from Found but not for lost...
I do not know what the solution is, can someone explain / give it to me?
Thanks in advance,
LTKort

Answer 1

Put the Matched IS NULL checks in the ON conditions of the LEFT JOIN, not WHERE.
SELECT vv.Airport_name,
COUNT(DISTINCT gb.Found_id) countFound,
COUNT(DISTINCT vb.Lost_id) countLost,
COUNT(DISTINCT kt.Match_id) countMatch
FROM Airports vv
LEFT JOIN Found gb ON vv.Airport_id = gb.Found_AirportID AND gb.Matched IS NULL
LEFT JOIN Lost vb ON vv.Airport_id = vb.Lost_AirportID AND vb.Matched IS NULL
LEFT JOIN Match kt ON vv.Airport_id = kt.Match_AirportID
GROUP BY vv.Airport_name
The problem with doing it in WHERE is that you're only getting the results where either the Lost or Found item was matched.

Answer 2

Alternatively, consider joining derived tables of aggregates to avoid many-to-many joins during the COUNT() evaluations:
SELECT a.AirportName, ftbl.countFound, lbtl.countLost, mtbl.countMatched
FROM Airports a
LEFT JOIN
(SELECT f.Found_AirportID, COUNT(f.Found_id) AS countFound
FROM Found f
WHERE f.Matched IS NULL
GROUP BY f.Found_AirportID) As ftbl
ON a.Airport_id = ftbl.Found_AirportID
LEFT JOIN
(SELECT l.Lost_AirportID, COUNT(l.Lost_id) AS countLost
FROM Lost l
WHERE l.Matched IS NULL
GROUP BY l.Lost_AirportID) As ltbl
ON a.Airport_id = ltbl.Lost_AirportID
LEFT JOIN
(SELECT m.Match_AirportID, COUNT(m.Match_id) AS countMatched
FROM Matched m
GROUP BY m.Match_AirportID) As mtbl
ON a.Airport_id = mtbl.Match_AirportID