Question

我有一个主表，它不包含这些列（Visit_date，M-1，M-2，M-3），但最终结果我需要这些列（Visit_date，M-1，M-2），M-3）以及现有的主表格列，例如

Retailer_id, survey_id, Group_Id, survery_Name

我尝试使用数据透视但是无法完成最终结果，请使用示例

来提示

主表：

 RetailerID Retailer SurveyName    Date     Score   Weights
  198760     ABC      Quality  06/03/2016   10       10
  198760     ABC      Quality  06/02/2016   5        10
  198760     ABC      Quality  06/01/2016   5        5
  198760     ABC      Quality  06/12/2015   10       10

预期结果：

Retailer_id  survery_Name    Last_Visit    M-1    M-2    M-3   Weightage
    198760     quality          10          5       5     10     10

脚本：

CREATE TABLE [dbo].[Master_Table]
(
    [Retailer_ID] [nvarchar](50) NULL,
    [RQSC_Survey_Name] [nvarchar](50) NULL,
    [RQSC_Date] [date] NULL,
    [RQSC_Weightage] [decimal](18, 0) NULL,
    [RQSC_Score] [decimal](18, 0) NULL
) ON [PRIMARY]   

INSERT [dbo].[Master_Table] ([Retailer_ID], [RQSC_Survey_Name], [RQSC_Date], [RQSC_Weightage], [RQSC_Score]) 
VALUES (N'198760', N'Quality', CAST(0x1C3B0B00 AS Date), CAST(10 AS Decimal(18, 0)), CAST(10 AS Decimal(18, 0)))

INSERT [dbo].[Master_Table] ([Retailer_ID], [RQSC_Survey_Name], [RQSC_Date], [RQSC_Weightage], [RQSC_Score]) 
VALUES (N'198760', N'Quality', CAST(0x003B0B00 AS Date), CAST(10 AS Decimal(18, 0)), CAST(5 AS Decimal(18, 0)))

INSERT [dbo].[Master_Table] ([Retailer_ID], [RQSC_Survey_Name], [RQSC_Date], [RQSC_Weightage], [RQSC_Score]) 
VALUES (N'198760', N'Quality', CAST(0xE13A0B00 AS Date), CAST(5 AS Decimal(18, 0)), CAST(5 AS Decimal(18, 0)))

INSERT [dbo].[Master_Table] ([Retailer_ID], [RQSC_Survey_Name], [RQSC_Date], [RQSC_Weightage], [RQSC_Score]) 
VALUES (N'198760', N'Quality', CAST(0xC23A0B00 AS Date), CAST(10 AS Decimal(18, 0)), CAST(10 AS Decimal(18, 0)))

我的查询：

SELECT
    Retailer_ID, RQSC_Survey_Name,
    [1] As LastVist, [2] as 'M-1', [3] as 'M-2', [4] as 'M-3',
    [1] as 'Score' 
FROM
    (SELECT 
         RQSC_Score, Retailer_ID, RQSC_Survey_Name, 
         RQSC_Weightage,
         ROW_NUMBER() OVER (PARTITION BY Retailer_ID ORDER BY RQSC_Date DESC) AS Rownumber 
     FROM 
         master_table 
     WHERE
         Retailer_ID = 198760 AND RQSC_Survey_Id = 298) src   
PIVOT(SUM(RQSC_Score)for Rownumber in ([1], [2], [3],[4])) piv;

如果我选择权重列，我会根据权重列获得重复行，但在预期结果中我应该得到权重列

Answer 1

对不起，我已经忙于工作。基于您的示例数据，我们可以实现使用Stuff和Dynamic Sql

IF object_id('tempdb..#Temp') is not null
    DROP TABLE #TEMP


CREATE  TABLE  #Temp
    (Retailer varchar(10), SurveyName varchar(10), Date datetime, Score int, Weights int)
;

INSERT INTO #Temp
    (Retailer, SurveyName, Date, Score, Weights)
VALUES
    ('198760', 'Quality', '2016-06-03 05:30:00', 10, 10),
    ('198760', 'Quality', '2016-06-02 05:30:00', 5, 10),
    ('198760', 'Quality', '2016-06-01 05:30:00', 5, 5),
    ('198760', 'Quality', '2015-06-12 05:30:00', 10, 10)
;



DECLARE @statement NVARCHAR(max)
,@columns NVARCHAR(max),
@col NVARCHAR(max)


SELECT @columns =
  STUFF((SELECT distinct '],[' +
                        CAST(ROW_NUMBER() OVER (PARTITION BY Retailer ORDER BY Date DESC) AS VARCHAR(50)) AS Rownumber 
                      FROM #Temp
                      FOR XML PATH(''), TYPE
                     ).value('.', 'NVARCHAR(MAX)') 
                        , 1, 2, '')+ ']'

SELECT @statement = 'SELECT
    Retailer, SurveyName,
    MAX([1]) As LastVist, MAX([2]) as ''M-1'', MAX([3]) as ''M-2'', MAX([4]) as ''M-3'',
    MAX([1]) as ''Score'' 
FROM
    (
    SELECT 
         Retailer, SurveyName, Score, 
         Weights,
         ROW_NUMBER() OVER (PARTITION BY Retailer ORDER BY Date DESC) AS Rownumber 
     FROM 
         #Temp

        ) src   
PIVOT(SUM(Score)for Rownumber in  (' + @columns + ')) as pvt
GROUP BY Retailer, SurveyName'

EXEC sp_executesql @statement = @statement

Answer 2

我得到了解决方案

WITH CTE as 
(

select retailer ,surveyname ,max([date]) 
over(partition by retailer,surveyname)as [date],score ,weights,
row_number() over(partition by retailer,surveyname order by [date] desc
)

as rn from #temp)  

select p.retailer,p.surveyname,p.[1] as lastvisit,p.[2] as [m-1],
p.[3] as [m-2],p.[4] as [m-3],b.score,b.weights ,b.[date] from 
(

select retailer,surveyname,score,rn from cte) as a  pivot (sum(score) 
for rn in ([1],[2],[3],[4])
) as p         
inner join cte as b on b.retailer=p.retailer and b.surveyname=p.surveyname    and b.rn=1  ;

在SQL Server中使用pivot动态构造静态列

2 个答案: