Question

我有两个表：

工作人员-持有员工ID，受雇公司，合同种类（比如说“ a”或“ b”）

+----+------------+----------+
| id | company_id | contract |
+----+------------+----------+
|  1 |          1 | a        |
|  2 |          1 | b        |
|  3 |          3 | b        |
|  4 |          2 | a        |
+----+------------+----------+

公司-保存公司ID和公司名称。

+----+--------------+
| id | company_name |
+----+--------------+
|  1 | Company 1    |
|  2 | Company 2    |
|  3 | Company 3    |
|  4 | Company 4    |
+----+--------------+

我想做的是选择在两种合同中雇用的每个公司的平均雇员数。这是我写的：

SELECT
  (SELECT COUNT(*)
   FROM Workers
   WHERE `contract` = 'a') / (SELECT COUNT(*)
                                FROM Companies) 'Contract A',
  (SELECT COUNT(*)
   FROM Workers
   WHERE `contract` = 'b') / (SELECT COUNT(*)
                               FROM Companies)  'Contract B';

还有：

SELECT
  sum(case when `contract` = 'a'
    then 1
      else 0 end) / (SELECT COUNT(*)
                     FROM Companies) 'Contract A',
  sum(case when `rodzaj contract` = 'uop'
    then 1
      else 0 end) / (SELECT COUNT(*)
                     FROM Companies) 'Contract B'
FROM Workers;

哪个效率更高？有没有更简单的方法？我正在使用MySQL。

预期输出为合同A-0,5，合同B-0,5，因为这4家公司的合同A中有2个人，而这4家公司的合同B中有2个人

Answer 1

我会去

SELECT SUM(contract = 'a') / c.cnt as contract_a,
       SUM(contract = 'b') / c.cnt as contract_b
FROM Workers w CROSS JOIN
     (SELECT COUNT(*) as cnt FROM Companies) c;

或者，如果您可以在单独的行中处理结果：

select contract, count(*) / cnt
from Workers w cross join
     (select count(*) as cnt from Companies) c
group by contract, cnt;

关于性能，将count(*)子句放在FROM子句中可确保只对它进行一次评估。因此，这似乎可以提高性能。

此外，仅对字符串和日期常量使用单引号。尝试使用不需要转义的列别名。

嵌套SELECT计算两个表的平均值

1 个答案: