2017-07-27 63 views
0

使用红移。我有一个具有以下字段的表:需要创建一个在日期前填写值的表格

Column:   Type: 
department | varchar    
employee_ID | varchar   
event   | varchar  
date   | date 

和只是有一个字段,并列出所有日期的日期表。 每个部门都有几个部门有员工。 “事件”字段有两个可能的值:JOIN或LEAVE。加入记录是他们加入公司/部门的日期,并且LEAVE记录是他们离开公司/部门的日期。它看起来像这样:

department employee_id event date 
marketing 001   JOIN 6/17/2017 
marketing 002   JOIN 6/19/2017 
marketing 002   LEAVE 6/20/2017 
marketing 001   LEAVE 6/22/2017 

我想使已经每天上市(再次我有所有日期以及表)表以及何时使用它们的值1和0的值当他们不是全体员工时。这应该是这样的:

date  department employee_id employed 
6/17/2017 marketing 001   1 
6/18/2017 marketing 001   1 
6/19/2017 marketing 001   1 
6/19/2017 marketing 002   1 
6/20/2017 marketing 001   1 
6/20/2017 marketing 002   1 
6/21/2017 marketing 001   1 
6/22/2017 marketing 001   1 

想我可能需要创建具有开始日期和同一行中离开日期一个新表?因为我可以使用case语句和分区来获取第一个和最后一个日期,但不知道该从哪里去......也许是一个理货表?该查询可能是这样,但仍难道不得到期望的结果:

select 
department, 
employee_id, 
CASE WHEN ROW_NUMBER() OVER (PARTITION BY employee_id ORDER BY date ASC) = 1 THEN date 
ELSE NULL 
END AS join_date, 
CASE WHEN ROW_NUMBER() OVER (PARTITION BY employee_id ORDER BY date DESC) = 1 THEN date 
ELSE NULL 
END AS leave_date 
from table1 
+0

即时通讯使用redshift – Berra2k

回答

-1

您需要使用您的CROSS JOIN表日期和包含加入和离开日期表之间。

仅限数据示例。你会使用你现有的表格。

CREATE TEMP TABLE dim_date (dt_val) 
AS  SELECT '2017-06-16'::date dt_val 
UNION ALL SELECT '2017-06-17'::date dt_val 
UNION ALL SELECT '2017-06-18'::date dt_val 
UNION ALL SELECT '2017-06-19'::date dt_val 
UNION ALL SELECT '2017-06-20'::date dt_val 
UNION ALL SELECT '2017-06-21'::date dt_val 
UNION ALL SELECT '2017-06-22'::date dt_val 
UNION ALL SELECT '2017-06-23'::date dt_val 
; 
CREATE TEMP TABLE empl_event (department, employee_id, event, event_dt) 
AS  SELECT 'marketing' department, 1 employee_id, 'JOIN' event, '2017-06-17'::date event_dt 
UNION ALL SELECT 'marketing' department, 2 employee_id, 'JOIN' event, '2017-06-19'::date event_dt 
UNION ALL SELECT 'marketing' department, 2 employee_id, 'LEAVE' event, '2017-06-20'::date event_dt 
UNION ALL SELECT 'marketing' department, 1 employee_id, 'LEAVE' event, '2017-06-22'::date event_dt 
; 

逻辑步骤1:将JOINLEAVE事件日期上的1行

CREATE TEMP TABLE empl_period (department, employee_id, join_dt, leave_dt) 
AS 
SELECT department 
     ,employee_id 
     ,MAX(CASE WHEN event = 'JOIN' THEN event_dt ELSE NULL END) join_dt 
     ,MAX(CASE WHEN event = 'LEAVE' THEN event_dt ELSE NULL END) leave_dt 
FROM empl_event 
GROUP BY 1,2 
; 

逻辑第2步:使用CROSS JOIN创建一排每一个可能的雇员日期组合和设置employed 1如果日期在join_dtleave_dt之间。

SELECT emp.department 
     ,emp.employee_id 
     ,ddt.dt_val 
     ,CASE WHEN ddt.dt_val BETWEEN emp.join_dt AND emp.leave_dt THEN 1 ELSE 0 END employed 
FROM  empl_period emp 
CROSS JOIN dim_date ddt 
ORDER BY 1,2,3 
; 

例如数据输出

department | employee_id | dt_val | employed 
------------+-------------+------------+---------- 
marketing |   1 | 2017-06-16 |  0 
marketing |   1 | 2017-06-17 |  1 
marketing |   1 | 2017-06-18 |  1 
marketing |   1 | 2017-06-19 |  1 
marketing |   1 | 2017-06-20 |  1 
marketing |   1 | 2017-06-21 |  1 
marketing |   1 | 2017-06-22 |  1 
marketing |   1 | 2017-06-23 |  0 
marketing |   2 | 2017-06-16 |  0 
marketing |   2 | 2017-06-17 |  0 
marketing |   2 | 2017-06-18 |  0 
marketing |   2 | 2017-06-19 |  1 
marketing |   2 | 2017-06-20 |  1 
marketing |   2 | 2017-06-21 |  0 
marketing |   2 | 2017-06-22 |  0 
marketing |   2 | 2017-06-23 |  0 
+0

真正的文件是方式t哟大到手动输入该工会的每个日期..还有其他事情我可以做吗? – Berra2k

+0

'dim_date'和'empl_event'仅为演示创建示例数据。没有必要写任何工会。 –

-1

您可以使用下面的SQL使用

DECLARE @dateranges TABLE (employee_id VARCHAR(4), 
          department VARCHAR(20), 
          join_date DATE, 
          leave_date DATE) 

INSERT @dateranges SELECT employee_id, 
          department, 
          MIN(event_date) join_date, 
          MAX(event_date) leave_date 
        FROM employment 
        GROUP BY employee_id, 
          department; 

WITH cte (dt, dept, id, emd) AS 
(
    SELECT tbl.join_date AS dt, 
      tbl.department AS dept, 
      tbl.employee_id AS employee_id, 
      1 AS emd 
    FROM @dateranges tbl 
    UNION ALL 
    SELECT DATEADD(DAY, 1, cte.dt) AS dt, 
      tbl.department AS dept, 
      tbl.employee_id AS employee_id, 
      1 AS emp 
    FROM cte 
    INNER JOIN @dateranges tbl 
     ON cte.id = tbl.employee_id 
     AND cte.dept = tbl.department 
    WHERE cte.dt < tbl.leave_date 
) 

SELECT dt AS date, 
     dept AS department, 
     id AS employee_id, 
     emd AS employed 
    FROM cte 
ORDER BY dt, id 

结果和日期

日期以扩大:

date department employee_id employed 
2017-06-17 marketing 001 1 
2017-06-18 marketing 001 1 
2017-06-19 marketing 001 1 
2017-06-19 marketing 002 1 
2017-06-20 marketing 001 1 
2017-06-20 marketing 002 1 
2017-06-21 marketing 001 1 
2017-06-22 marketing 001 1 
+0

这看起来很棒谢谢你!有一个问题,我被告知一个交叉连接函数会做到这一点,但似乎并没有在红移中可用。你有任何想法,如果交叉加入是红移的运营商,或者如果有其他选择? – Berra2k

+0

此SQL不适用于Redshift。 –

相关问题