Как запросить количество по 2 столбцам

Question

Я пытаюсь написать SQL-запрос, который показывает, как часто две команды играли друг против друга.

Id | Team1 | Team2 | Date 
1  |   A   |   B   | 25/5/11
2  |   B   |   A   | 26/5/11
3  |   A   |   C   | 27/5/11
4  |   C   |   B   | 28/5/11
5  |   A   |   B   | 28/5/11

результат должен быть:

A vs B => 3
A vs C => 1
C vs B => 1 

Подсчет AB и BAтак как это простой запрос.Но я не могу заставить их считаться вместе.

Есть предложения?

Answer 1

Я использовал подзапрос, чтобы изменить порядок команд перед группировкой.

SELECT first_team, second_team, count(*)
FROM (
    SELECT 
        CASE WHEN Team1 < Team2 THEN Team1 ELSE Team2 END AS first_team,
        CASE WHEN Team1 < Team2 THEN Team2 ELSE Team1 END AS second_team
    FROM table
) a
GROUP BY first_team, second_team;

Answer 2

SELECT Team1, Team2, SUM(num) FROM (
   SELECT Team1, Team2, COUNT(*) num
   FROM table_name
   GROUP BY Team1, Team2
  UNION ALL
   SELECT Team2, Team1, COUNT(*) num
   FROM table_name
   GROUP BY Team2, Team1
) combined
WHERE Team1 < Team2
GROUP BY Team1, Team2

Редактировать: обновлено для изменения команд при необходимости.

Примечание. Это будет выполняться намного быстрее, чем версии, использующие CASE, которые вы указали в других ответах, поскольку в нем будут полностью использованы индексы.

Edit2: перемещено, где быть еще быстрее с индексами.

Answer 3

Есть несколько способов достижения вашей цели:

SELECT Teams, Games = COUNT(*) FROM 
(
  SELECT 
    Teams = CASE WHEN Team1 < Team2 THEN Team1 ELSE Team2 END + ' vs ' +
            CASE WHEN Team1 < Team2 THEN Team2 ELSE Team1 END
  FROM MY_TABLE 
) AS T
GROUP BY Teams

ИЛИ, если вы используете SQL 2005/2008

;WITH T AS (
  SELECT 
    Teams = CASE WHEN Team1 < Team2 THEN Team1 ELSE Team2 END + ' vs ' +
            CASE WHEN Team1 < Team2 THEN Team2 ELSE Team1 END
  FROM MY_TABLE 
)
SELECT Teams, Games = COUNT(*) FROM T GROUP BY Teams

Оба из вышеперечисленных даст вам тот же результат

/*
Teams     Games
-------|------
A vs B  3
A vs C  1
B vs C  1
*/

Вот сценарий, с которым вы можете играть:

/* TEST DATA */
DECLARE @t AS TABLE ( ID INT, Team1 CHAR(1), Team2 CHAR(1), playdate [DATETIME] )
INSERT INTO @t (Team1 , Team2 , playdate)
          SELECT 'A' , 'B', '20110525'
UNION ALL SELECT 'B' , 'A', '20110526'    
UNION ALL SELECT 'A' , 'C', '20110527'    
UNION ALL SELECT 'C' , 'B', '20110528'    
UNION ALL SELECT 'A' , 'B', '20110528'    


/* STYLE 1 */    
;WITH T AS (
  SELECT 
    Teams = CASE WHEN Team1 < Team2 THEN Team1 ELSE Team2 END + ' vs ' +
            CASE WHEN Team1 < Team2 THEN Team2 ELSE Team1 END
  FROM @t 
)
SELECT Teams, Games = COUNT(*) FROM T GROUP BY Teams

/* STYLE 2 */
SELECT Teams, Games = COUNT(*) FROM 
(
  SELECT 
    Teams = CASE WHEN Team1 < Team2 THEN Team1 ELSE Team2 END + ' vs ' +
            CASE WHEN Team1 < Team2 THEN Team2 ELSE Team1 END
  FROM @t 
) AS T
GROUP BY Teams

/ * ИЛИ, используйте его для переключения столбцов * /

/* STYLE 3 */    
;WITH T AS (
  SELECT 
      Team1 = CASE WHEN Team1 < Team2 THEN Team1 ELSE Team2 END 
    , Team2 = CASE WHEN Team1 < Team2 THEN Team2 ELSE Team1 END
  FROM @t 
)
SELECT Team1 , Team2, Games = COUNT(*) FROM T GROUP BY Team1 , Team2

/* STYLE 4 */
SELECT Team1 , Team2, Games = COUNT(*) FROM 
(
  SELECT 
      Team1 = CASE WHEN Team1 < Team2 THEN Team1 ELSE Team2 END 
    , Team2 = CASE WHEN Team1 < Team2 THEN Team2 ELSE Team1 END
  FROM @t 
) AS T
GROUP BY Team1 , Team2

Answer 4

SELECT 
  (CASE WHEN Team1<Team2 THEN Team1 ELSE Team2) Team1,
  (CASE WHEN Team1>Team2 THEN Team1 ELSE Team2) Team2,
  COUNT(*)  cnt
FROM <table> 
GROUP BY 
  (CASE WHEN Team1<Team2 THEN Team1 ELSE Team2) Team1,
  (CASE WHEN Team1>Team2 THEN Team1 ELSE Team2) Team2