在MySQL查询中对记录的子集执行一些算术运算

Question

I have a MySQL table containing attendance records. Relevant columns identify attendees, sessions and attendance status (confirmed/declined/no-response). I also have a second table containing attendees' personal info, including the date they joined the group.

I am trying to build a single query that will output a list of all attendees, the number of times they attended sessions, and their regularity, expressed as the percentage of sessions they attended (against the total number of sessions). The problem is that total of sessions should only contains the sessions since the attendee joined the group.

So, for the first part, my query is:

SELECT name, count(*)
FROM attendance, members
WHERE attendance.attend = 2 and
      attendance.member_number = members.number
GROUP BY attendance.member_number, attendance.attend
HAVING count(*) > 0
ORDER BY count(*) desc

This returns all attendees' names, plus the number of sessions each had attended. However, I'd also like to add that third column.

Thus, for each record, I first have to find out what was the members.member_since date, then query only the group of attendance records for each member, for which the dates are greater than member_since, get their count, then from that subset get the count where atteneance.attend = 2, then calculate what percentage of that subset count represents the second count (of records having attend = 2).

How do I write my query?

Data looks like this:

Table members:

number, name, join_date
001, Peter James, 2016-07-26
002, John Smith, 2014-06-04

TABLE attendance:

member_number, attend, date
001, 2, 2019-05-11
001, 1, 2019-05-04
001, 2, 2019-04-26
001, 2, 2019-04-18
002, 2, 2014-08-01
002, 1, 2016-08-01
002, 1, 2019-05-04

The desired result would be to get the following:

Name, Attended, percentage
Peter James, 3, 75%
John Smith, 1, 14%

In other words, while there were 6 sessions, Peter James has only been member for the last four, and attended 3 of them (75%), while John Smith has been a member since the beginning, so his attendance is one out of 6.

The structure is somewhat simplified (there is a 'sessions' table with details about sessions, including the date, and the 'attendance' table refers to its key column, 'att_number', rather than the actual date), but the point is the same.

Answer 1

Join the tables just on the dates to get the total number of sessions since the member joined. Then use a conditional SUM() to get the number of sessions that the member attended. Divide them to get the percentage.

SELECT m.name, SUM(m.number = a.member_number AND a.attend = 2) AS attendance, 100*SUM(m.number = a.member_number AND a.attend = 2)/COUNT(*) AS percent
FROM members AS m
JOIN attendance AS a ON a.date >= m.member_since
GROUP BY m.number
ORDER BY attendance DESC