Question

我使用的是SQL Server 2016，我有一张包含以下记录的表：

Create table MemberCount 
(
    ssn char(9) null,
    extract_begin_dt smalldatetime null,
    extract_end_dt   smalldatetime null,
    memberCase varchar(13) null
)

insert into MemberCount(ssn, extract_begin_dt, extract_end_dt, memberCase) 
values (1, '1/1/2017', '1/31/2017', 'Non-Funded'),
       (1, '2/1/2017', '12/31/2017', 'OpenAccts'),
       (2, '1/1/2017', '3/31/2017', 'OpenAccts'),
       (2, '4/1/2017', '12/31/2017', 'OpenAccts'),
       (3, '1/1/2014', '3/31/2014', 'OpenAccts'),
       (3, '4/1/2014', '10/31/2017', 'ClosedAccts'),
       (3, '11/1/2017', '12/31/2017', 'OpenAccts'),
       (7, '9/19/2017', '12/31/2017', 'OpenAccts'),
       (7, '9/15/2017', '9/18/2017', 'OpenAccts'),
       (7, '8/26/2017', '9/14/2017', 'OpenAccts'),
       (7, '4/9/2015', '8/25/2017', 'OpenAccts'),
       (7, '4/8/2015', '4/8/2015', 'OpenAccts'),
       (7, '4/7/2015', '4/7/2015', 'ClosedAccts'),
       (7, '1/1/2015', '4/6/2015', 'OpenAccts'),
       (7, '5/22/2014', '12/31/2014', 'OpenAccts'),
       (7, '8/21/2013', '5/21/2014', 'OpenAccts'),
       (7, '6/20/2013', '8/20/2013', 'Non-Funded'),
       (7, '4/20/2013', '6/19/2013', 'Non-Funded'),
       (7, '1/12/2011', '1/12/2011', 'DeletedAccts'),
       (7, '12/31/2010', '1/11/2011', 'OpenAccts')

预期结果

ssn   begin_dt
----------------
1     2/1/2017
2     1/1/2017
3     11/1/2017
7     12/31/2017

规则是如果没有memberCase，那么OpenAccount将选择最小日期，如果有其他那么OpenAccount它将一直运行直到开放账户将是当前的

我尝试过使用Windows功能但是数字2和3是相反的逻辑

谢谢，

Oded Dror

Answer 1

检查此查询。但结果与您提供的结果略有不同。对于ssn=7，您期望12/31/2017。但它是extract_end_dt。你没有得到extract_begin_dt吗？

select
    ssn, extract_begin_dt
from (
    select 
        *, mc1 = min(memberCase) over (partition by ssn), mc2 = max(memberCase) over (partition by ssn)
        , dt1 = min(extract_begin_dt) over (partition by ssn), dt2 = max(extract_begin_dt) over (partition by ssn)
    from 
        MemberCount
) t
where
    1 = case
        when mc1 = mc2 and mc1 = 'OpenAccts' and dt1 = extract_begin_dt then 1
        when mc1 <> mc2 and dt2 = extract_begin_dt then 1
        else 0
    end

结果

ssn        extract_begin_dt
1          2017-02-01 00:00:00
2          2017-01-01 00:00:00
3          2017-11-01 00:00:00
7          2017-09-19 00:00:00

编辑：根据您的评论更新查询

select
    ssn, begin_dt = max(begin_dt)
from (
    select
        ssn, grp, begin_dt = min(extract_begin_dt) 
    from (
        select
            *, grp = rn - row_number() over (partition by ssn order by extract_begin_dt)
        from (
            select
                *, rn = row_number() over (partition by ssn order by extract_begin_dt)
            from
                MemberCount
        ) t
        where
            memberCase = 'OpenAccts'
    ) t
    group by ssn, grp
) t
group by ssn

最新类别

1 个答案: