mysql query - multiple counts using left join and where clause - mysql

Question

I'm currently trying to get the following data:
UserName, UserImageURL, Total Games Played, Games Completed, Games Lost, Average Won (as percentage) and Points of the user
And as well another set of data:
User Statistics data such as:
Most Games Played on League: 23 - Monster Killers
Games Most Won On: 19/23 - Monster Killers
Games Most Lost On: 3/32 - Frog Racers
Your Game Winning Accuracy (total from all games) - 68% accuracy
Site Stats:
Most Games Played on League: 650 - Helicopter Run
Top Game Played: 1200 - Monster Killers
Whole site winning accuracy: 82%
I have the following Tables:
-User Table-
userID (int-pk), userName (varchar), userImageUrl (text)
-Games table-
gameId (int-pk), gameName (varchar), gameUserID (int), gameLeagueId (int), score1 (int), score2 (int), gameResultOut (0 or 1), gameWon (0 or 1)
-UserBalance table-
ubId(int-pk) userId (int) balance (int)
-League table-
leagueId (int-pk) leagueName (varchar)
Just to give you a heads up on what's happening, when a user plays a game and chooses some results a row is inserted into the games table. Since the game is time based, when the results are out, there is a check that checks if there are any games which have that id and will update the gameResultOut to 1 and gameWon to 1 or 0 according to what the user had selected as a score.
I tried the following:
SELECT u.userID, u.userName, u.userImageUrl, l.leagueName ,
COUNT(g.gameId) AS predTotal,
(SELECT COUNT(g.gameId) FROM games AS g WHERE g.gameResultOut = 1 AND g.gameWon = 1) AS gamesWon,
(SELECT COUNT(g.gameId) FROM games AS g WHERE g.gameResultOut = 1 AND g.gameWon = 0) AS gamesLost,
ub.balance
FROM games AS g
LEFT JOIN league AS l ON l.leagueId = g.gameLeagueId
LEFT JOIN user AS u ON u.user_id = g.gameUserID
LEFT JOIN user_balance AS ub ON ub.userId = u.userID
WHERE l.leagueId = 4
GROUP BY u.userId
ORDER BY ub.balance DESC
I can calculate easily the win percentage after the query so that's not a problem, but the result for the Wins and Lost are all the same and even when it comes to changing the leageId, the results are still the same which is not what I want.
Can anyone help?
Thanks & Regards,
Necron

Answer 1

As far as I see, the games table stores games that users played. So, in order to know how many games each user played/won/lost, you're missing the link in the subqueries between games and users.
Your subqueries are:
(SELECT COUNT(g.gameId ) FROM games AS g WHERE g.gameResultOut = 1 AND g.gameWon = 1) AS gamesWon,
(SELECT COUNT(g.gameId) FROM games AS g WHERE g.gameResultOut = 1 AND g.gameWon = 0) AS gamesLost,
And they should be:
(SELECT COUNT(gw.gameId ) FROM games AS gw WHERE gw.gameResultOut = 1 AND gw.gameWon = 1 AND gw.gameUserID = u.user_id) AS gamesWon,
(SELECT COUNT(gl.gameId) FROM games AS gl WHERE gl.gameResultOut = 1 AND gl.gameWon = 0 AND gl.gameUserID = u.user_id) AS gamesLost,
I guess this is what you're looking for :)
EDIT based on comments, adding tips for User and Site statistics:
For those information you'll need to perform several distinct queries, as most of them are going to sum some values and/or group by a given column, which won't fit for another query. I'll try to give you some ideas so you can work on them.
User Statistics
Most Games Won or Lost
The previous answer for the query you provided counts how many times user has lost/won any game, but does not distinct this data between games.
So, if you want to know in which game user has most wins/losses, you should have something like this:
SELECT
g.gameName,
-- How many times the user won per game
(SELECT COUNT(gw.gameId) FROM games gw WHERE gw.gameResultOut = 1 AND gw.gameWon = 1 AND gw.gameUserID = u.user_id) AS gamesWon,
-- How many times the user payed each game
COUNT(g.gameId) AS gamesPlayed,
-- The Win Ratio. This may need a little work on, depending on what you want.
-- Be aware that if a user played a game 1 time and won, it's ratio will be 1 (100%)
-- So maybe you'll want to add your own rule to determine which game should show up here
(gamesWon / gamesPlayed) AS winRatio
FROM
games g
INNER JOIN user u ON u.user_id = g.gameUserID
-- Groups and counts data based on games + users
GROUP BY g.gameId, u.user_id
-- Now you order by the win ratio
ORDER BY winRatio DESC
-- And get only the first result, which means the game the player has most wins.
LIMIT 1
For lost games, it's pretty much the same query, changing the desired fields and maths.
Game winning accuracy
Somewhat the previous query, except that you won't group by the gameID anymore. Just group by the user and do your math.
Site Statistics
Well, as far as I see, we're still on a similar query. The difference is that for the whole Site statistics you won't ever group by user. You may group by game or league, depending on what you are trying to achieve.
Bottom line: looks that most queries are similar, you'll have to play with them and adapt for each information you need to retrieve. Please note that they might not work plenty as I could not test them on your DB. You may need to correct some inconsistence according to your database/tables schema.
I hope this may give you some insight to work on.