Put Condition to CASE expression (SQL) - sql

Question

Is there anyway that I can filter the customers whose total bets on in-play football is more than 50% of their total bets everywhere? This is what I have in mind.
There are 2 tables, where the sportsbook_bets contains all bets from all customers (a customer may place multiple bets).
Sample Data:
Customer table
customer_id
1
2
...
sportsbook_bet table
customer_id
sport_name
1
football
1
football
1
hockey
1
basketball
2
tennis
2
football
2
hockey
2
basketball
...
...
Based on the tables above, the query should return customer 1 since the total bets placed on football is at least 50% of the total bet everywhere; where customer 2 only have 25% (thus doesn't get queried).
SELECT c.customer_id
FROM customers c INNER JOIN sportsbook_bets s
ON c.customer_id=s.customer_id
WHERE 50 < ROUND((SUM(CASE WHEN s.sport_name = 'football' AND s.in_play_yn='Y' THEN 1 ELSE 0 END)/COUNT(s.bet_id))*100,0);

Answer 1

When performing a filter using an aggregate, need to use HAVING not WHERE.
Customers with >= 50% Bets on Football
SELECT c.customer_id
FROM customers c
INNER JOIN sportsbook_bets s
ON c.customer_id=s.customer_id
GROUP BY c.customer_id
HAVING COUNT(CASE WHEN s.sport_name = 'football' AND s.in_play_yn='Y' THEN 1 END)/1.0/COUNT(*) >= .50
If you want to see a row per bet for those the desired customers, could use this query instead:
WITH cte_CustomerBets AS (
SELECT
c.customer_id
,s.sports_name
/*Add any other columns you need as well*/
,COUNT(CASE WHEN s.sport_name = 'football' AND s.in_play_yn='Y' THEN 1 END) OVER (PARTITION BY CustomerID) AS TotalSportsBetsPerCustomer
,COUNT(*) OVER (PARTITION BY CustomerID) AS TotalBetsPerCustomer
FROM customers c
INNER JOIN sportsbook_bets s
ON c.customer_id=s.customer_id
GROUP BY c.customer_id
HAVING COUNT(CASE WHEN s.sport_name = 'football' AND s.in_play_yn='Y' THEN 1 END)/1.0/COUNT(*) >= .50
)
SELECT *
FROM cte_CustomerBets
WHERE TotalSportsBetsPerCustomer/1.0/TotalBetsPerCustomer >= .5

Answer 2

I'm going to assume that you need some other columns from your customer table, otherwise you don't need that table to get customer_id, as that column is already in sportsbook_bets table.
One solution is to use a subqueries:
select distinct
C.customer_id
From customers C
Join sportsbook_bets S On S.customer_id = C.customer_id
where (Select count(FB.bet_id) From sportsbook_bets FB where FB.customer_id = C.customer_id and FB.sport_name = 'football'and FB.in_play_yn = 'Y') >
((Select count(TB.bet_id) From sportsbook_bets TB where TB.customer_id = C.customer_id) / 50.0)
Other way that I like more it's to separate your calculations, for this you can use the APPLY operator:
Select distinct
C.customer_id
From customers C
Join sportsbook_bets S On S.customer_id = C.customer_id
Cross Apply
(
Select TotalBets = count(SB.bet_id)
From sportsbook_bets SB
where SB.customer_id = S.customer_id
) TB
Cross Apply
(
Select FootballBets = count(SB.bet_id)
From sportsbook_bets SB
where SB.customer_id = S.customer_id
and SB.sport_name = 'football'
and SB.in_play_yn = 'Y'
) FB
where FB.FootballBets > TB.TotalBets / 50.0

Answer 3

For readability, you can separate the two components -- total bets and in-play football bets -- into two queries. This may not perform as quickly as other solutions here, but it may be easier to understand and maintain.
WITH totalbets AS (
SELECT c.customer_id
, count(s.bet_id) as TotalBets
FROM customers c
INNER JOIN sportsbook_bets s ON c.customer_id = s.customer_id
GROUP BY c.customer_id
),
ipfbets AS (
SELECT customer_id
, count(bet_id) as IPFBets
FROM sportsbook_bets
WHERE sport_name = 'football'
and in_play_yn = 'Y'
GROUP BY customer_id
)
SELECT tb.customer_id
FROM totalbets tb
INNER JOIN ipfbets ipfb ON tb.customer_id = ipfb.customer_id
WHERE (1.0 * ipfb.IPFBets) / tb.TotalBets > 0.5