Question

我正在寻找一种方法来容纳有多个结果来找到最大值和最小值。我找到了上一个问题的链接： max Counts

给出的答案之一是：

SELECT  color_id, COUNT(color_id) totalCount
FROM    products 
WHERE   item_id = 1234 
GROUP   BY color_id 
HAVING  COUNT(color_id) =
(
  SELECT  COUNT(color_id) totalCount
  FROM    products 
  WHERE   item_id = 1234 
  GROUP   BY color_id 
  ORDER BY totalCount DESC
  LIMIT 1  
)

这是否是大型数据库特别适用的做法？如果这有意义，上述查询本身是否基本上在自身内运行？

我有一个更复杂的查询，还需要找到ma和min。我想优化它：

修改：

SELECT `system_users`.`first`, `system_users`.`last`,  COUNT(`quotes`.`created_by`) as most_quotes
FROM `quotes` 
INNER JOIN `system_users`
ON `quotes`.`created_by` = `system_users`.`id`
where `system_users`.`store_id` = '$createdID' 
and `quotes`.`date_created` between '$startDate' and '$endDate' group by(`created_by`)
HAVING count(`quotes`.`created_by`) = 
(
SELECT COUNT(`quotes`.`created_by`)
FROM `quotes` 
INNER JOIN `system_users`
ON `quotes`.`created_by` = `system_users`.`id`
where `system_users`.`store_id` = '$createdID' 
and `quotes`.`date_created` between '$startDate' and '$endDate' group by(`created_by`) ORDER BY count(`created_by`) DESC limit 1
)
OR
(
SELECT COUNT(`quotes`.`created_by`)
FROM `quotes` 
INNER JOIN `system_users`
ON `quotes`.`created_by` = `system_users`.`id`
where `system_users`.`store_id` = '$createdID'
and `quotes`.`date_created` between '$startDate' and '$endDate' group by(`created_by`) ORDER BY count(`created_by`) ASC limit 1
)
ORDER BY most_quotes ASC

到目前为止，我正试图想出找到最大和最小的不同方法而没有运气。对此更多的帮助将不胜感激谢谢 MC

Answer 1

这是一个坏主意 - 在大型数据库上使用HAVING。而且，此外，您的问题可以通过这种方式解决（我有MySQL 5.5版本）：

SELECT  
  color_id, 
  COUNT(color_id) AS totalCount
FROM    
  products 
WHERE   
  item_id = 1234 
GROUP BY 
  color_id 
ORDER BY 
  totalCount DESC 
LIMIT 1

HAVING的问题在于它在整个查询完成后执行，即存储引擎已经工作，因此不能对HAVING条件进行索引或其他优化 - 因此，它可以被视为完整的结果集扫描。

感谢@GordonLinoff，我发现它并不是你想要的。如果你想找到所有相应的行，你最好像戈登建议的那样行事。

虽然我找到了解决此问题的另一种方法，但它可能只比使用HAVING的原始变体好一些（并且 - 更好，因为存储引擎将同时涉及）

SELECT
  first.color_id, 
  first.rows_count 
FROM 
  (SELECT color_id, COUNT(1) AS rows_count FROM products WHERE item_id=1234 GROUP BY color_id) AS first 
  LEFT JOIN
  (SELECT color_id, COUNT(1) AS rows_count FROM products WHERE item_id=1234 GROUP BY color_id ORDER BY rows_count DESC LIMIT 1) AS second 
    ON first.rows_count=second.rows_count 
WHERE second.rows_count IS NOT NULL;

我也有变量变体（类似于戈登的变量）。所以你可以选择这些选项。

Answer 2

您的问题的答案是，有多种方法可以使用一个查询在MySQL中处理此问题。

一种方法是问题中的方法。另一种方法是使用变量：

select color_id, totalcount
from (SELECT color_id, COUNT(color_id) as totalCount,
             @maxcount = if(@maxcount < count(color_id), count(color_id), @maxcount)
      FROM products cross join
           (select @maxcount := 0) const
     WHERE item_id = 1234 
     GROUP BY color_id 
    ) t
where totalCount = @maxcount;

我不是在MySQL中使用变量的忠实粉丝，但通常它们是有效表达查询的唯一方式。我相信这是将您想要的内容写成单个查询的最有效方式。

第三种方法是使用临时表。第四种方法是将having条件中的子查询移动到另一个连接。

在大多数其他数据库中，您只需使用窗口函数。类似的东西：

select color_id, totalcount
from (SELECT color_id, COUNT(color_id) as totalCount,
             max(count(color_id)) over () as maxTotalCount
      FROM products
     WHERE item_id = 1234 
     GROUP BY color_id 
    ) t
where totalCount = maxTotalCount;

但是，MySQL不支持这些。

编辑：

如果你想使用变量的最大值和最小值：

select color_id, totalcount
from (SELECT color_id, COUNT(color_id) as totalCount,
             @maxcount = if(@maxcount < count(color_id), count(color_id), @maxcount),
             @mincount = if(@mincount > count(color_id) or @mincount = -1, count(color_id), @mincount)
      FROM products cross join
           (select @maxcount := -1, @mincount := -1) const
     WHERE item_id = 1234 
     GROUP BY color_id 
    ) t
where totalCount in (@mincount, @maxcount);

mysql Max有多个实例

2 个答案: