根据 ROW_NUMBER 高于 1 的结果进行过滤

Question

我设法使用 ROW_NUMBER() 获得了一些有序的结果，并按国家代码和客户参考号进行了分区，并按报价日期排序。 ROW_NUMBER() 列称为 PRank。

CountryCode  ¦   ClientRef   ¦   Quote Ref   ¦   Name     ¦    Address     ¦   PostCode   ¦   PreferencesCode   ¦   QuoteDate         ¦   PolicyType   ¦   PRank   ¦
     0           AABB001         AABB001MB1    Mr A Smith   1 Country Lane     WA21 2PU       5934611             2017-03-18 00:00:00     MB               1
     0           AABB001         AABBMM1MB2    Mr A Smith   1 Country Lane     WA21 2PU       5934611             2017-03-18 00:00:00     MB               2
     0           AABB001         AABBMM1MB3    Mr A Smith   1 Country Lane     WA21 2PU       5934611             2017-03-18 00:00:00     MB               3
     0           BBGG003         BBGG003MB1    Mrs B Jones  2 City Road        M1 3XY         312191              2017-09-02 00:00:00     MB               1

这是我的 ROW_NUMBER() 专栏的代码：

ROW_NUMBER() OVER(PARTITION BY CountryCode, ClientRef, QuoteDate 
                  ORDER BY QuoteDate DESC, PolicyRef asc)

有很多像最后一行这样的结果；他们只计为 1，因为客户在相关日期只有一个报价。我想消除这些行，而不从客户结果中删除在同一日期出现多个报价的任何行（即我仍然希望在上面的结果中看到第一个客户的所有三行）所以我想通过使用 CTE 和过滤在 ROW_NUMBER 上会导致我丢失我仍然需要查看的结果中的数据。

我试图用 HAVING 做一些事情并对 PRank 列求和，但我无法弄清楚分组，因为我不想实际聚合数据，只需删除 ClientRef 只有的任何行每天一个 QuoteRef。

Answer 1

如果您正在寻找一天内有超过 1 个报价的客户，CTE 仍然是最好的选择。

假设这与 InfocentrePlus 中的 ic_dapolicy 视图相反（这显然是 OpenGI 数据，并且 icp 表不那么容易使用）那么您的查询将类似于：

WITH Quotes AS(
    SELECT *,
           COUNT(*) OVER (PARTITION BY PPY.B@, PPY.Ref@) AS ClientQuotes 
    FROM ic_Dapolicy PPY
    WHERE PPY.Quote_date = @Quote_date) --If you aren't filtering on Quote_date in your where, add it to the PARTITION BY clause
SELECT [YourColumns]
FROM Quotes
WHERE ClientQuotes > 1;

Answer 2

我想你想要像下面这样的东西。

;WITH Rankings AS
(
    -- Your query here
    SELECT
        CountryCode = 1,
        ClientRef = 1,
        QuoteDate = 1,
        PRank = 1
),
Only1Record AS
(
    SELECT
        R.CountryCode, 
        R.ClientRef, 
        R.QuoteDate
    FROM
        Rankings AS R
    GROUP BY
        R.CountryCode, 
        R.ClientRef, 
        R.QuoteDate
    HAVING
        MAX(R.PRank) = 1
)
DELETE D FROM
    Only1Record AS R
    INNER JOIN Rankings AS D ON
        R.CountryCode = D.CountryCode AND
        R.ClientRef = D.ClientRef AND
        R.QuoteDate = D.QuoteDate

另一个更简单的变体是使用窗口COUNT()。

;WITH Rankings AS
(
    -- Your query here
    SELECT
        CountryCode = 1,
        ClientRef = 1,
        QuoteDate = 1,
        PRank = 1,
        Total = COUNT(1) OVER (PARTITION BY CountryCode, ClientRef, QuoteDate)
)
DELETE D FROM
    Rankings AS D
WHERE
    D.Total = 1

根据 ROW_NUMBER 高于 1 的结果进行过滤

Filter based on results where ROW_NUMBER goes higher than 1

sql

sql-server

filtering

row-number