变更号码栏

Question

我有NAME和PAY，但在此示例中我需要CHANGEGROUP：

NAME   PAY  DATE    CHANGEGROUP
Sally   12  10/01/2011  1
Sally   12  10/01/2011  1
Sally   12  11/02/2011  1
Sally   12  11/02/2011  1
Sally   12  12/01/2012  1
Sally   13  04/23/2013  2
Sally   12  04/24/2013  3
Sally   10  05/01/2013  4
Sally   10  10/01/2014  4

我尝试了RANK()和DENSE_RANK()，但是它们根据值进行了分组-
因为薪水下降了，所以弄乱了我的分组。我看到了这一点，但它与此旧版本的SQL 2005不兼容

Answer 1

这是一个鸿沟和孤岛的问题。

一种方法。SQL小提琴

WITH T1
     AS (SELECT *,
                ROW_NUMBER()
                  OVER (
                    PARTITION BY NAME
                    ORDER BY DATE) - ROW_NUMBER()
                                       OVER (
                                         PARTITION BY NAME, [PAY]
                                         ORDER BY DATE) AS Grp
         FROM   Table1),
     T2
     AS (SELECT *,
                MIN(DATE)
                  OVER (
                    PARTITION BY NAME, Grp) AS MinDate
         FROM   T1)
SELECT [NAME],
       [PAY],
       [DATE],
       DENSE_RANK()
         OVER (
           PARTITION BY NAME
           ORDER BY MinDate) AS CHANGEGROUP
FROM   T2
ORDER  BY NAME,
          MinDate