MySQL: Count Duplicate Rows in Multiple Tables Without Inflating Counts - mysql

Question

I'm attempting to count how many completed events each person in my table have done. The problem I'm running into is that people have multiple jobs in my person table, which means there are purposeful multiple rows per person -- which is making my event table double when I try to do counts.
Here's a SQL Fiddle of my code. Easiest to see ID #1 only has two events completed but the table counts four because they also have two jobs.
Here's my sample schema:
CREATE TABLE persontable
(id INT NOT NULL
, name VARCHAR(255) NOT NULL
, employer VARCHAR(255) NOT NULL
, PRIMARY KEY(id,employer)
);
CREATE TABLE eventtable
(id INT NOT NULL
, name VARCHAR(255) NOT NULL
, eventname VARCHAR(255) NOT NULL
, eventdate DATE NOT NULL
, status VARCHAR(255) NOT NULL
, PRIMARY KEY (id,eventname,eventdate));
INSERT INTO persontable (id,name,employer) VALUES
(1,"Joe","Party Inc."),
(1,"Joe","Body Shop"),
(2,"Puddy","Body Shop"),
(3,"Newman","Postal Service"),
(3,"Newman","Computers Inc."),
(4,"Delores","Mulva LLC"),
(5,"Morty","Executive Raincoats"),
(6,"Helen","Body Shop"),
(7,"Frank","Retired"),
(7,"Frank","Mulva LLC"),
(8,"Estelle","Retired"),
(9,"Mandelbaum","Weight Lifters Guild"),
(9,"Mandelbaum","The Wiz"),
(10,"Fred","The Wiz");
INSERT INTO eventtable (id,name,eventname,eventdate,status) VALUES
(1,"Joe","Mayo Party",5/4/94,"Completed"),
(1,"Joe","Coat Shopping",1/2/95,"Completed"),
(4,"Delores","Play",5/9/94,"Completed"),
(4,"Delores","Name Guessing",3/9/98,"Completed"),
(9,"Mandelbaum","Working Out",3/2/97,"Declined"),
(10,"Fred","Store Sale",8/9/96,"Completed");
And my fairly simple query that's adding the additional counts:
SELECT
p.id,
e.id,
COUNT(DISTINCT CASE WHEN e.status="Completed" THEN e.id ELSE NULL END) AS EVENT,
COUNT(CASE WHEN e.status="Completed" THEN e.id ELSE NULL END) AS YTDAllShiftsComp
FROM persontable p
LEFT JOIN eventtable e ON p.id = e.id
GROUP BY p.id;
My desired outcome for the sample is:
id id EVENT YTDAllShiftsComp
1 1 1 2
2 (null) 0 0
3 (null) 0 0
4 4 1 2
5 (null) 0 0
6 (null) 0 0
7 (null) 0 0
8 (null) 0 0
9 9 0 0
10 10 1 1
Thanks for the help!

Answer 1

Thats what happens when you dont normalize your data.Since each person can attend multiple events and each event can host multiple persons you need an intermediate table which holds the primary keys of both tables,this is called many to many relation.So I Joined just on distinct persons id,eliminating the duplicates,but the real solution is to add a new table.
SELECT
x.id,
e.id,
COUNT(DISTINCT CASE WHEN e.status="Completed" THEN e.id ELSE NULL END) AS EVENT,
COUNT(CASE WHEN e.status="Completed" THEN e.id ELSE NULL END) AS YTDAllShiftsComp
FROM (SELECT id FROM persontable GROUP BY id)x
LEFT JOIN eventtable e ON x.id = e.id
GROUP BY x.id;

Answer 2

You can use correlated subqueries:
SELECT
p.id,
(SELECT COUNT(DISTINCT CASE WHEN e.status="Completed" THEN e.id END)
FROM eventtable e
WHERE p.id = e.id) AS EVENT,
(SELECT COUNT(CASE WHEN e.status="Completed" THEN e.id END)
FROM eventtable e
WHERE p.id = e.id) AS YTDAllShiftsComp
FROM persontable p
GROUP BY p.id;
Demo here

Answer 3

As Georgios mentioned, you need subqueries - but if that sometimes-null 2nd ID column is really needed, you'll want to wrap the main statement to NULL it out if the event count is zero.
SELECT id, if(event=0, NULL, event) as idagain, event, ytdallshiftscomp
FROM (SELECT distinct p.id,
(SELECT count(distinct id) FROM eventtable WHERE id=p.id AND status="Completed") AS EVENT,
(SELECT count(*) FROM eventtable WHERE id=p.id AND status="Completed") AS ytdallshiftscomp
FROM persontable p) q