SQL Server 返回出现不同的记录答案

【问题标题】：SQL Server return records with different occurrenceSQL Server 返回出现不同的记录
【发布时间】：2019-02-26 23:22:41
【问题描述】：

如下表。

SupplierId  AuditId  QuestionId  Answer
3           123      1           abc
3           456      1           abc
3           789      1           abc

我想看看在这张表中的什么地方发生了以下情况

SupplierId  AuditId  QuestionId  Answer
3           123      1           abc
3           456      1           abc
3           789      1           def

我想知道D 列偏离的位置。我在想像ROW_NUMBER这样的窗口函数

SELECT SupplierId, AuditId, QuestionId, Answer , ROW_NUMBER() OVER (PARTITION BY SupplierId, AuditId, QuestionId ORDER BY SupplierId) AS numOccurences
FROM Analytics2.Data.Question
ORDER BY SupplierId
OFFSET 0 ROWS
FETCH NEXT 100 ROWS ONLY;

【问题讨论】：

这些答案之一是否解决了您的问题？如果没有，您能否提供更多信息来帮助回答？否则，请考虑标记最能解决您的问题的答案（上/下投票箭头下的复选标记）。见stackoverflow.com/help/someone-answers

标签： sql sql-server tsql

【解决方案1】：

以下查询将为您提供所有那些在其中具有多个不同 D 列值的供应商 ID

    SELECT SupplierId, count(distinct D) 
      AS numOccurences
       FROM Analytics2.Data.Question
           group by SupplierId having 
          count(distinct D) >1
         ORDER BY SupplierId

【讨论】：

【解决方案2】：

我认为您应该从 Partition By 子句中删除 AuditID：

SELECT SupplierId, 
       AuditId, 
       QuestionId, 
       Answer , 
       ROW_NUMBER() OVER (PARTITION BY SupplierId,  QuestionId 
                          ORDER BY SupplierId,QuestionId) AS numOccurences
FROM Analytics2.Data.Question
ORDER BY sipplierId

如果您只需要跟踪更改，请使用公用表表达式：

WITH CTE_1 AS (
    SELECT SupplierId, 
       AuditId, 
       QuestionId, 
       Answer , 
       ROW_NUMBER() OVER (PARTITION BY SupplierId,  QuestionId 
                          ORDER BY SupplierId,QuestionId) AS numOccurences
FROM Analytics2.Data.Question

)SELECT * FROM CTE_1 
 WHERE numOccurences = 1
 ORDER BY sipplierId

【讨论】：

【解决方案3】：

此查询将为您提供表中Answer 字段更改的行（和行号）：

WITH cte AS (
    SELECT *,
           ROW_NUMBER() OVER (ORDER BY SupplierId) AS row
    FROM Question)
SELECT cte2.* 
FROM cte cte1
JOIN cte cte2 ON cte2.row = cte1.row + 1 AND cte2.Answer != cte1.Answer

输出：

SupplierId  AuditId     QuestionId  Answer  row
3           789         1           def     3

Demo on dbfiddle

【讨论】：

【解决方案4】：

你没有说你使用的是什么版本的 SQL Server。 SQL Server 2012 实现了LEAD 和LAG 函数。它们对于此类查询非常有效。

这是如何使用它们的一种变体。调整 PARTITION BY 和 ORDER BY 子句以符合您的要求，这在问题中并不清楚。

WITH 
CTE
AS
(
    SELECT 
        SupplierId, AuditId, QuestionId, Answer
        ,LAG(Answer) OVER (PARTITION BY SupplierId ORDER BY AuditID) AS PrevAnswer
    FROM Analytics2.Data.Question
    ORDER BY SupplierId, AuditID
)
SELECT
    SupplierId, AuditId, QuestionId, Answer
FROM CTE
WHERE
    Answer <> PrevAnswer
;

【讨论】：