Question

假设我有一张能够按月记录就业状况的表格。我有一个场景，一名员工从实习生开始，回到学校并全职回来。

理想情况下，我希望沿着一条线输出，这几乎是他/她的工作经历的历史。

date_hire    date_termination    employee_id    employee_type
2012_01_01   2012_03_01          10002          intern
2012_05_01   null                10002          full-time

数据看起来像

date_snapshot    date_hire    date_termination    employee_id    employee_type
2012_01_01       2012_01_01   null                10002          intern
2012_02_01       2012_01_01   null                10002          intern
2012_03_01       2012_01_01   2012_03_01          10002          intern
2012_04_01       2012_01_01   2012_03_01          10002          intern
2012_05_01       2012_05_01   null                10002          full-time
2012_06_01       2012_05_01   null                10002          full-time
2012_07_01       2012_05_01   null                10002          full-time
......

我尝试使用GROUP BY技术

SELECT
    date_hire,
    date_termination,
    employee_id,
    employee_type
FROM
    employees
GROUP BY
    date_hire,
    date_termination,
    employee_id,
    employee_type

给我回复，请注意我有一个不需要的记录

date_hire    date_termination    employee_id    employee_type
2012_01_01   null                10002          intern
2012_01_01   2012_03_01          10002          intern
2012_05_01   null                10002          full-time

Answer 1

如果我们确定员工无法返回之前的类型（例如实习生 - > AWF - >全职 - > AWF），那么有一个简单的解决方案：

select      date_hire
           ,min (date_termination)  as date_termination
           ,employee_id
           ,employee_type

from        employees e

group by    date_hire
           ,employee_id
           ,employee_type
;

否则 -

select    date_hire
         ,date_termination
         ,employee_id
         ,employee_type

from     (select    e.*
                   ,lead (date_hire)        over (partition by e.employee_id order by date_snapshot) as next_date_hire
                   ,lead (date_termination) over (partition by e.employee_id order by date_snapshot) as next_date_termination
                   ,lead (employee_type)    over (partition by e.employee_id order by date_snapshot) as next_employee_type
                   ,max  (date_snapshot)    over (partition by e.employee_id)                        as max_date_snapshot

          from      employees e
          )

where     date_hire         <> next_date_hire
      or  date_termination  <> next_date_termination
      or  employee_type     <> next_employee_type
      or  date_snapshot     =  max_date_snapshot
;

Answer 2

以下查询还应提供您想要的输出。

SELECT DISTINCT MIN(date_hire) OVER (PARTITION BY employee_id,employee_type) date_hire,
       MIN(date_termination)  OVER (PARTITION BY employee_id,employee_type) date_termination,
      employee_type,employee_id
FROM employees;

通过每周快照查找就业历史

2 个答案: