SQL查询返回最常见，第二最常见，第三最常见的情况

Question

我可以使用数据透视表在Excel中完成此操作，但我想找出一种直接从单个SQL查询中执行此操作的方法。 假设我有一份水果及其新鲜度的清单：

**Fruit   Freshness**
Banana  New
Banana  Ripe
Apple   Ripe
Orange  Old
Cherry  New
Orange  Ripe
Apple   Old
Banana  Old
Apple   New
Apple   New
Orange  Ripe
Banana  Old
Orange  New
Cherry  New
Cherry  Ripe

我想计算“新鲜度”出现的次数，然后将它们排列为最频繁，第二最频繁，依此类推。 结果看起来像这样：

**Fruit Most common 2nd most common 3rd most common**
Banana  New         Old             Ripe
Apple   Old         New             Ripe
Orange  Ripe        New             Old
Cherry  New         Ripe            NA

一个查询有可能吗？

Answer 1

您可以只使用GROUP_CONCAT（）聚合函数：

SELECT
  Fruit, GROUP_CONCAT(Freshness ORDER BY cnt DESC) as Common
FROM (
  SELECT Fruit, Freshness, COUNT(*) cnt
  FROM
    fruits
  GROUP BY
    Fruit, Freshness
) s

这将返回如下值：

Fruit  | Common
---------------------
Banana | New,Old,Ripe
Cherry | New,Ripe
...    | ...

但是，如果要将结果分为三列，则可以合并上一个查询并使用SUBSTRING_INDEX（）从逗号分隔的值中提取第一个，第二个和第三个值：

SELECT
  Fruit,
  SUBSTRING_INDEX(Common, ',', 1) AS most,
  CASE WHEN CommonLIKE '%,%'
       THEN SUBSTRING_INDEX(SUBSTRING_INDEX(Common, ',', 2), ',', -1) END AS second_most,
  CASE WHEN CommonLIKE '%,%,%'
       THEN SUBSTRING_INDEX(SUBSTRING_INDEX(Common, ',', 3), ',', -1) END AS third_most
FROM (
  SELECT
    Fruit, GROUP_CONCAT(Freshness ORDER BY cnt DESC) as Common
  FROM (
    SELECT Fruit, Freshness, COUNT(*) cnt
    FROM
      fruits
    GROUP BY
      Fruit, Freshness
  ) s
  GROUP BY
    Fruit
  ) s

Answer 2

SQL Server 2008及更高版本：

select fruit, [1], [2], [3] from
( select row_number() over (partition by fruit order by ct desc) as rn, fruit, freshness from (
select count(1) as ct,  fruit, freshness from f
group by fruit, freshness ) g ) src
PIVOT
(
MAX(Freshness)
FOR rn in ([1], [2], [3])
) pvt

Answer 3

尝试这个

create table #fr (Fruit  varchar(20), Freshness varchar(20))
insert #fr values
('Banana' , 'New'),('Banana' , 'Ripe'),('Apple'  , 'Ripe'),('Orange' , 'Old'),('Cherry' , 'New'),
('Orange' , 'Ripe'),('Apple'  , 'Old'),('Banana' , 'Old'),('Apple'  , 'New'),('Apple'  , 'New'),
('Orange' , 'Ripe'),('Banana', 'Old'),('Orange' , 'New'),('Cherry',  'New'),('Cherry',  'Ripe')


SELECT Fruit,
       [1] Most_Common,
       [2] Second_Common,
       [3] Third_common
FROM   (SELECT Fruit,Freshness,
               Row_number()OVER(partition BY Fruit ORDER BY Count(*) DESC) rn
        FROM   #fr
        GROUP  BY Fruit,Freshness) a
       PIVOT (Max(Freshness)
             FOR rn IN([1],[2],[3])) piv