如何优化子查询

Question

我有一个包含多个子查询的查询，用于根据状态计算某个交易中涉及的用户数量。

SELECT * FROM (
    SELECT
    DATE(changes.created_at) AS `date`,
    SUM(IF(changes.status = 15, 1, 0)) AS installed_daily,
    SUM(IF(changes.status = 3, 1, 0)) AS port_reserved,
    (
      SELECT COUNT(DISTINCT created_by) FROM applicant_state_changes
      WHERE DATE(created_at) = DATE(changes.created_at) AND status = 3
    ) AS user_port_reserved,
    SUM(IF(changes.status = 5, 1, 0)) AS document_validated,
    (
      SELECT COUNT(DISTINCT created_by) FROM applicant_state_changes
      WHERE DATE(created_at) = DATE(changes.created_at) AND status = 5
    ) AS user_document_validated,
    SUM(IF(changes.status = 7, 1, 0)) AS account_created,
    (
      SELECT COUNT(DISTINCT created_by) FROM applicant_state_changes
      WHERE DATE(created_at) = DATE(changes.created_at) AND status = 7
    ) AS user_account_created,
    SUM(IF(changes.status = 11, 1, 0)) AS jo_created,
    (
      SELECT COUNT(DISTINCT created_by) FROM applicant_state_changes
      WHERE DATE(created_at) = DATE(changes.created_at) AND status = 11
    ) AS user_jo_created
    FROM applicant_state_changes AS changes
    GROUP BY DATE(changes.created_at)
    LIMIT 100 OFFSET 0
) a
ORDER BY date ASC;

这大约需要 130 秒。没有子查询，我的查询最多只需要 0.5 秒。

Answer 1

您可以使用 case when 表达式进行条件聚合

 SELECT
    DATE(changes.created_at) AS `date`,
    SUM(IF(changes.status = 15, 1, 0)) AS installed_daily,
    SUM(IF(changes.status = 3, 1, 0)) AS port_reserved,
    count(distinct case when status = 3 then created_by end) as user_port_reserved
    ,
    SUM(IF(changes.status = 5, 1, 0)) AS document_validated,
    count(distinct case when status = 5 then created_by  end) AS user_document_validated,
    SUM(IF(changes.status = 7, 1, 0)) AS account_created,
    count(distinct case when status = 7 then 1 end) AS user_account_created,
    SUM(IF(changes.status = 11, 1, 0)) AS jo_created,
    count(distinct case when status = 11 then created_by  end) AS user_jo_created
    FROM applicant_state_changes AS changes where 
    GROUP BY DATE(changes.created_at)
    LIMIT 100 OFFSET 0

如何优化子查询

How to optimize subquery

mysql

select

count

distinct