Same queries giving different results - sql

Question

So for an assignment at school, we had to extract a count from a database. The question was as follows,
--19) How many airports in a timezone name containing 'Europe' are used as a source airport (source_airport_id) for a route that has aircraft that have a wake of 'M' or 'L'
This was the code I came up with,
SELECT count(DISTINCT airports.id) FROM airports WHERE timezone_name LIKE '%Europe%' AND id IN
(SELECT source_airport_id FROM routes WHERE id IN
(SELECT id FROM route_aircrafts WHERE aircraft_id IN
(SELECT id FROM aircrafts WHERE wake_size IN ('M', 'L'))));
it returned 544, while the professors answer returned 566.
SELECT count (DISTINCT airports.id)
FROM airports, routes, route_aircrafts, aircrafts
WHERE airports.id = routes.source_airport_id
AND routes.id = route_aircrafts.route_id
AND aircrafts.id = route_aircrafts.aircraft_id
AND airports.timezone_name LIKE'%Europe%'
AND aircrafts.wake_size IN ('M', 'L'); --566
To me, those two should be doing the same thing and I can't understand why the answers are different.

Answer 1

To get the same answer in your query you need:
SELECT count(DISTINCT airports.id) FROM airports WHERE timezone_name LIKE '%Europe%' AND id IN
(SELECT source_airport_id FROM routes WHERE id IN
(SELECT route_id FROM route_aircrafts WHERE aircraft_id IN
(SELECT id FROM aircrafts WHERE wake_size IN ('M', 'L'))));
You'd used the primary ID field rather than the foreign key route_id. You were getting an approximately similar result because there must be a significant overlap in the values.

Answer 2

I would go with something along the lines of:
SELECT COUNT(DISTINCT airports.id)
FROM airports
INNER JOIN routes ON airports.id = routes.source_airport_id
INNER JOIN route_aircrafts ON routes.id = route_aircrafts.route_id
INNER JOIN aircrafts ON route_aircrafts.aircraft_id = aircrafts.id
AND aircrafts.wake_size IN ('M', 'L')
WHERE airports.timezone_name LIKE '%Europe%'
Explanation:
SELECT COUNT(DISTINCT airports.id)
You don't want to count duplicate airports.ids more than once.
FROM airports
This is the main table you're counting from. All other tables build from this one.
INNER JOIN routes ON airports.id = routes.source_airport_id
INNER JOIN will only include rows that match in both tables. Matching on airports.id and routes.source_airport_id.
INNER JOIN route_aircrafts ON routes.id = route_aircrafts.route_id
INNER JOIN will only include rows that match in both tables. Matching on routes.id and route_aircrafts.route_id.
INNER JOIN aircrafts ON route_aircrafts.aircraft_id = aircrafts.id
AND aircrafts.wake_size IN ('M', 'L')
Same thing with the INNER JOINs above. We've added an additional filter for wakes. For an INNER JOIN, this filter can also be performed in the WHERE clause without changing the results. Putting filters in the JOIN keeps the intent together (and the optimizer will likely filter this way anyway). For an OUTER JOIN, filtering in the JOIN vs filtering in the WHERE can possibly return different results (depending on your data).
WHERE airports.timezone_name LIKE '%Europe%'
Now we are filtering the entire resultset by the timezone_name from the base table of airports.
When working with SQL, it's important to think of your data in SETS. This will help you write more performant, and less programatic, queries.