Question

Let’s have an example to declare my concern:
Suppose we have a Table (Tags) which has two columns like this

UserID -------------------------------- Tag
1 -------------------------------------- SQL
1 -------------------------------------- Select
1 -------------------------------------- DB
2 -------------------------------------- SQL
2 -------------------------------------- Programming
2 -------------------------------------- Code
2 -------------------------------------- Software
3 -------------------------------------- Code
4 -------------------------------------- SQL
4 -------------------------------------- Code

I need to count DISTINCT co-occurrences for each tag based on UserID

So, the output should be like this (with Order by Co-occurrences desc):

Tag -------------------------------- Co-occurrences
---------------------------------------------
SQL --------------------------------------- 5
Programming ------------------------------- 3
Code -------------------------------------- 3
Software ---------------------------------- 3
Select ------------------------------------ 2
DB ---------------------------------------- 2

This is just an example..

How can I make a Select statement that can do this?

I came up with one way but for only ONE specific tag:

SELECT count (distinct (Tag)) - 1 as Co_occurrences
FROM Tags
WHERE Tag is NOT NULL and UserID in
    (   SELECT UserID
        FROM Tags 
        where tag = 'SQL')

Is it possible to change the above statement to make it general for all tags in the table?

Answer 1

SELECT t2.tag, count (distinct (t1.Tag)) - 1 as Co_occurrences
FROM Tags t1 inner join 
    Tags t2 on t1.UserId = t2.UserId 
GROUP BY t2.tag    
ORDER BY count (distinct (t1.Tag)) desc