Question

我的数据与此示例类似，非常庞大，因此理想情况下需要高效的代码。我想在充电器维修后找到那些只有那些维修的客户ID的交易。

    TRANSACTION_ID  REPAIR_DATE CUSTOMER_ID    COMPONENT   LABOR_CODE_DESC Size....
    28289           6/25/2015   AH123        LAPTOP       CHARGER REPAIR   big   
    28235           6/29/2015    AH123       LAPTOP         CHIP REPLACE small
    258978         6/27/2013      HW687      PHONE      TOUCH SCREEN
    28223           6/2/2014     AH123       LAPTOP      BATTERY REPAIR
    215678          7/28/2014       HW687      PHONE         SIM REPAIR
    527808          7/30/2016      HW687      LAPTOP         BATTERY REPAIR
    567976          7/28/2014      HW687      LAPTOP         CHARGER REPAIR  big
    7678698         8/68/2015     AH123        LAPTOP        BATTERY REPAIR
    9987908          5/7/2006       TU890       PHONE      SIM REPAIR
    .....


    OUTPUT
   TRANSACTION_ID  REPAIR_DATE CUSTOMER_ID COMPONENT    LABOR_CODE_DESC ....
   28235           6/29/2015    AH123       LAPTOP         CHIP REPLACE
   7678698         8/68/2015      AH123        LAPTOP        BATTERY REPAIR
   527808          7/30/2016     HW687      LAPTOP          BATTERY REPAIR


   .....

dont need:
  215678          9/7/2014       HW687      PHONE     SIM REPAIR

因为它与充电器维修的日期相同。我尝试了以下代码

   SELECT *
   FROM tab
   QUALIFY 
   Max(CASE WHEN LABOR_CODE_DESC = 'CHARGER REPAIR' THEN 1 ELSE 0 END)
   Over (PARTITION BY CUSTOMER_ID
     ORDER BY REPAIR_DATE
     ROWS BETWEEN Unbounded Preceding AND 1 Preceding) >= 1

通过使用这个我错过了一些交易发生在同一天充电器维修日期可能是因为它按修理日期排序。我不妨忽略所有与充电器维修日期相同的交易，以避免此问题。我也想根据尺寸进行限制。我可以在哪里加入？请提出最有效的方法，因为我的桌子太大了。

Answer 1

这在Teradata中有效吗？

   SELECT *
   FROM tab
   QUALIFY REPAIR_DATE > Max(CASE WHEN LABOR_CODE_DESC = 'CHARGER REPAIR' THEN REPAIR_DATE
                             END) Over (PARTITION BY CUSTOMER_ID
                                        ORDER BY REPAIR_DATE
                                        ROWS BETWEEN Unbounded Preceding AND 1 Preceding
                                       );

这样可以在充电器维修之后的日期开始。

SQL查询在特定部分修复后查找所有事务

1 个答案: