How this query be optimised to prevent timeout - mysql

Question

From the database below with schema
movieActor (actorID, movieID)
rental (rentalID, inventoryID, customerID)
inventory (inventoryID, movieID)
I am trying to list pair of customers who rented movies from same actor. The resulting set should be composed of three columns
customerID1,customerID2,nOfCommonActors
for example
23 44 5
11 44 3
where the first row means customers with ids of 23 and 44 each rented various movies but 5 of those actors played in both set of movies customer 23 and 44 rented
I came up with this query however it takes so much time to run and times out without returning any result. Was wondering how I can make it more efficient( I am using MYSQL):
SELECT r1.customerID AS customerID1,
r2.customerID AS customerID2,
COUNT(DISTINCT fa.actorID) as nOfCommonActors
FROM movieActor AS fa
JOIN (SELECT r.customerID, i.movieID, fa.actorID
FROM rental AS r
JOIN inventory i
ON i.inventoryID=r.inventoryID
JOIN movieActor AS fa
ON fa.actorID=i.movieID
) AS r1
JOIN (SELECT r.customerID, i.movieID, fa.actorID
FROM rental AS r
JOIN inventory i
ON i.inventoryID=r.inventoryID
JOIN movieActor AS fa
ON fa.actorID=i.movieID
) AS r2
ON r2.actorID=r1.actorID
AND r1.customerID < r2.customerID
GROUP BY r1.customerID, r2.customerID
ORDER BY nOfCommonActors DESC;

Answer 1

The one thing I can think of is select distinct in the subqueries:
SELECT ca.customerID AS customerID1,
ca2.customerID AS customerID2,
COUNT(*) as nOfCommonActors
FROM (SELECT DISTINCT r.customerID, fa.actorID
FROM rental r JOIN
inventory i
ON i.inventoryID = r.inventoryID JOIN
movieActor fa
ON fa.actorID = i.movieID
) ca JOIN
(SELECT DISTINCT r.customerID, fa.actorID
FROM rental r JOIN
inventory i
ON i.inventoryID = r.inventoryID JOIN
movieActor fa
ON fa.actorID = i.movieID
) ca2
ON ca.actorID = ca2.actorID AND
ca.customerID < ca2.customerID
GROUP BY ca.customerID, ca2.customerID
ORDER BY nOfCommonActors DESC;
Your version is multiplying out the number of rows in the subqueries considerably. That makes the JOIN more expensive -- and all that extra work is for nought because you want COUNT(DISTINCT) anyway.

Answer 2

Splitting the query into, allows allows statistic to plot best path
SELECT DISTINCT r.customerID, fa.actorID
into #t1
FROM rental r JOIN
inventory i
ON i.inventoryID = r.inventoryID JOIN
movieActor fa
ON fa.actorID = i.movieID
SELECT DISTINCT r.customerID, fa.actorID
into #t1
FROM rental r JOIN
inventory i
ON i.inventoryID = r.inventoryID JOIN
movieActor fa
ON fa.actorID = i.movieID
select #t1.customerID AS customerID1,
#t2.customerID AS customerID2,
COUNT(*) as nOfCommonActors
from
(
select #t1.customerID, #t2.customerID
from #t1
join #t2 ON #t1.actorID = #t2.actorID AND #t1.customerID < #t2.customerID )
GROUP BY #t1.customerID, #t2.customerID
ORDER BY nOfCommonActors DESC;