MySQL interpreter calls query without an aggregate function an aggregate query - mysql

Question

I am trying to solve this HackerRank SQL coding challenge. We have two tables, one called Hackers with columns hacker_id and name and another called Submissions with columns submission_date, submission_id, hacker_id, and score.
One query I submitted to solve the challenge was:
SELECT es.date, es.count, m.hacker_id, m.name, m.score
FROM
(
SELECT submission_date as date, COUNT(hacker_id) as count
FROM (
SELECT submission_date, COUNT(submission_id) as count, hacker_id
FROM Submissions
GROUP BY submission_date, hacker_id
) f
HAVING count >=1
) es
JOIN (
SELECT s.submission_date as date, s.hacker_id, h.name, s.score
FROM Submissions s
JOIN Hackers h ON h.hacker_id = s.hacker_id
JOIN (SELECT submission_date, MAX(score) as score FROM Submissions GROUP BY submission_date) foo ON foo.submission_date = s.submission_date
WHERE s.score = foo.score
) m
ON es.date = m.date
ORDER BY es.date
However, I got back the error:
ERROR 1140 (42000) at line 1: In aggregated query without GROUP BY, expression #1 of SELECT list contains nonaggregated column 'f.submission_date'; this is incompatible with sql_mode=only_full_group_by
This really confused me since I didn't use an aggregate function in my query. Why did the interpreter give this error, and what should I think about to resolve it?

Answer 1

There are two columns named count in first subquery. One is count of field submission_id and second is count of hacker_id. As I understood query needs to perform selection of submission_dates where count greater than or equal to 1. So depends on which count is presumed to be greater than or equal to 1, the query might be rewritten as following:
1) If it is needed that count(submission_id) greater than or equal to 1:
SELECT es.date, es.count, m.hacker_id, m.name, m.score
FROM
(
SELECT submission_date as date, COUNT(hacker_id) as count
FROM (
SELECT submission_date, COUNT(submission_id) as count, hacker_id
FROM Submissions
GROUP BY submission_date, hacker_id
) f
WHERE f.count >=1
GROUP BY submission_date
) es
JOIN (
SELECT s.submission_date as date, s.hacker_id, h.name, s.score
FROM Submissions s
JOIN Hackers h ON h.hacker_id = s.hacker_id
JOIN (SELECT submission_date, MAX(score) as score FROM Submissions GROUP BY submission_date) foo ON foo.submission_date = s.submission_date
WHERE s.score = foo.score
) m
ON es.date = m.date
ORDER BY es.date
2) the count(hacker_id) is greater than or equal to 1:
SELECT es.date, es.count, m.hacker_id, m.name, m.score
FROM
(
SELECT submission_date as date, COUNT(hacker_id) as count
FROM (
SELECT submission_date, COUNT(submission_id) as count, hacker_id
FROM Submissions
GROUP BY submission_date, hacker_id
) f
GROUP BY submission_date
HAVING count(hacker_id) >= 1
) es
JOIN (
SELECT s.submission_date as date, s.hacker_id, h.name, s.score
FROM Submissions s
JOIN Hackers h ON h.hacker_id = s.hacker_id
JOIN (SELECT submission_date, MAX(score) as score FROM Submissions GROUP BY submission_date) foo ON foo.submission_date = s.submission_date
WHERE s.score = foo.score
) m
ON es.date = m.date
ORDER BY es.date
The problem is that in subquery the aggregated function count(hacker_id) was used without "group by" section with column "submission_date" defined there as nonaggregated column.
Still I'm not sure that the sql-query is logically correct.