MySQL - Count number of rows in each group

Question

I have a SQL table user_game which contains the games that a user owns:

| id | user_id | game_id |
|----|---------|---------|
| 83 | 1       | 1       |
| 84 | 1       | 2       |
| 85 | 1       | 3       |
| 86 | 2       | 2       |
| 87 | 2       | 3       |
| 88 | 2       | 4       |
| 89 | 3       | 2       |

I am trying to count the number of users which have 1 game, 2 games, 3 games.. etc.

User 1 has 3 games, User 2 has 3 games, and User 3 has 1 game. Therefore these are the results I want to achieve:

| no_of_games | COUNT(no_of_games) |
|-------------|--------------------|
| 1           | 1                  |
| 2           | 0                  |
| 3           | 2                  |

COUNT(no_of_games) is the number of users that have that number of games.

I can individually get the number of users for each no_of_games with this query:

-- Select no. of users with 1 game
SELECT no_of_games, COUNT(no_of_games)
FROM
(
  -- Select no. of games each user has
  SELECT user_id, COUNT(1) as no_of_games
  FROM user_game
  GROUP BY user_id
) as A
WHERE no_of_games = 1;

which gives the results:

| no_of_games | COUNT(no_of_games) |
|-------------|--------------------|
| 1           | 1                  |

However I have to change the no_of_games = 1 to 2, 3, 4... manually and UNION them with this solution and I can't do it for ~60 cases.

Is there a simpler way to achieve this?

Typically, you would handle the logic of missing results in application code- so, why not do that? — Strawberry, Jan 10 '17 at 07:34

Tim Biegeleisen · Accepted Answer · 2017-01-10T06:49:32.507

3

Your problem is a bit tricky, because groups of games which do not appear in your data with a certain frequency (e.g. 2) will not appear in the result set just using your original table. In the query below, I use a second table called nums which simply contains the sequence 1 through 10 representing counts of number of games. By using a LEFT JOIN we can retain each game count in the final result set.

SELECT t1.no_of_games,
       COALESCE(t2.no_of_games_count, 0) AS no_of_games_count
FROM nums t1
LEFT JOIN
(
    SELECT t.no_of_games, COUNT(*) AS no_of_games_count
    FROM
    (
        SELECT COUNT(*) AS no_of_games
        FROM user_game
        GROUP BY user_id
    ) t
    GROUP BY t.no_of_games
) t2
    ON t1.no_of_games = t2.no_of_games
ORDER BY t1.no_of_games

And here is the definition I used for nums:

CREATE TABLE nums (`no_of_games` int);
INSERT INTO nums (`no_of_games`)
VALUES
    (1),(2),(3),(4),(5),(6),(7),(8),(9),(10);

Demo here:

SQLFiddle

edited Jan 10 '17 at 06:49

answered Jan 10 '17 at 06:41

Tim Biegeleisen

502,043
27
286
360

Strange (I think) to use 10 instead of 0 - but whatever floats your boat, I guess! – Strawberry Jan 10 '17 at 07:36
@Strawberry My thinking is that a 0 entry would mean number of people who had zero games, but in this case it would mean the person didn't even appear to begin with. – Tim Biegeleisen Jan 10 '17 at 07:43
I get it, but to my mind, a 0-9 utility table is easily scalable. – Strawberry Jan 10 '17 at 07:51

score 1 · Answer 2 · answered Jan 10 '17 at 06:41

1

You can find count of games for each user and then find count of users for each count of games.

select cnt no_of_games, count(*) cnt_no_of_games
from(
    select user_id, count(*) cnt
    from your_table
    group by user_id
) t group by cnt;

answered Jan 10 '17 at 06:41

Gurwinder Singh

38,557
6
51
76

MySQL - Count number of rows in each group

2 Answers2

SQLFiddle