Question

我在SQL Server中有一个FormSummaries表，其中包含以下相关的示例数据列：

FormName | CompletionTime
Form1    | 70
Form1    | 20
Form1    | 30
Form1    | 40
Form1    | 80
Form1    | 60
Form1    | 90
Form1    | 10
Form2    | 30
Form2    | 40
Form2    | 80
Form2    | 90
Form2    | 40
Form2    | 1000
Form2    | 120
Form2    | 70

我需要做的是：

1）通过表单的名称和该表单的完成时间的平均值对数据进行分组，这很容易：

SELECT 
    FormName, AVG(CompletionTime) 
FROM 
    FormSummaries 
WHERE 
    CompletionTime  is not null
GROUP BY
    FormName

2）获得每种表单类型的完成时间的前25％/最低25％的平均值（即完成每个表单所花费的平均最快和最慢25％）。理想情况下，这将是一个查询，即

FormName | Bottom25%AverageCompletionTime | Top25%AverageCompletionTime
Form1    | 85                             | 15
Form2    | 560                            | 35

我生活在现实世界，并意识到这可能是不可能的，因此对于顶部和底部的单独查询将是正常的，即

FormName | Bottom25%AverageCompletionTime
Form1    | 85                            
Form2    | 560                           

FormName | Top25%AverageCompletionTime
Form1    | 15
Form2   | 35

我看过Partition by，Ntile and Over但是我似乎无法获得任何产生预期结果的东西（尽管这可能是因为我没有实现这些正确！）。

有人可以帮忙吗？

非常感谢。

Answer 1

NTILE以块的形式对结果进行排名，因此您对四分之一感兴趣，因此请使用NTILE（4）分成4组，并对formname进行分区。要使用2个查询执行此操作，请尝试

-- top 25%
SELECT  formname, AVG(CompletionTime) 
FROM
(SELECT 
    FormName,completiontime, NTILE(4) over (partition by FormName order by completiontime) as QuartPercentile
FROM 
    FormSummaries
WHERE CompletionTime IS NOT NULL )
    x
WHERE  QuartPercentile = 1
GROUP BY formname

-- bottom 25%
SELECT  formname, AVG(CompletionTime) 
FROM
(SELECT 
    FormName,completiontime, NTILE(4) over (partition by FormName order by completiontime) as QuartPercentile
FROM 
    FormSummaries 
WHERE CompletionTime IS NOT NULL)
    x
WHERE  QuartPercentile = 4
GROUP BY formname

或使用一个查询

SELECT  formname,AVG( case when QuartPercentile = 4 then CompletionTime else null end)   as [Bottom25%AverageCompletionTime]
, AVG( case when QuartPercentile = 1 then CompletionTime else null end)   as [Top25%AverageCompletionTime]
FROM
(SELECT 
    FormName,completiontime, NTILE(4) over (partition by FormName order by completiontime) as QuartPercentile
FROM 
    FormSummaries 
WHERE CompletionTime IS NOT NULL)
    x

GROUP BY formname

请记住，如果您的completiontime列有整数，AVG将返回一个整数，因此您可能希望转换以获得所需的精度，例如

AVG( case when QuartPercentile = 1 then cast(CompletionTime AS decimal(9,2))  else null end)

Answer 2

您可以使用CTE + PIVOT：

<xsl:stylesheet version="1.0"
    xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
    <xsl:output omit-xml-declaration="no" indent="yes"/>
    <xsl:strip-space elements="*"/>  

    <xsl:template match="node()|@*"> 
        <xsl:copy>
            <xsl:apply-templates select="node()|@*"/>
        </xsl:copy>
    </xsl:template>

    <xsl:template match="IconComponent[@Id='10001']//ColorId/text()[.='210']">21</xsl:template>

</xsl:stylesheet>

输出：

;WITH PercentCount AS (
SELECT  FormName,
        COUNT(*)/4 as [Bottom25Percent],
        COUNT(*) as [Top25Percent]
FROM Forms
GROUP BY FormName
), FormsWithRowNumber AS (
SELECT  f.FormName,
        f.CompletionTime,
        ROW_NUMBER() OVER (PARTITION BY f.FormName ORDER BY f.CompletionTime) as rn
FROM Forms f
), final AS (
SELECT  f.FormName, 
        f.CompletionTime,       
        CASE WHEN f.rn between 1 and [Bottom25Percent] THEN 1 
             WHEN f.rn between [Top25Percent]-[Bottom25Percent]+1 and [Top25Percent] THEN 2
             ELSE 0 END as [TopBottom]
FROM FormsWithRowNumber f
INNER JOIN PercentCount p
    ON p.FormName = f.FormName
)

SELECT *
FROM final
PIVOT (
    AVG(CompletionTime) FOR TopBottom IN ([1],[2])
) as pvt

SQL Server - 分组依据，平均值和百分位数

2 个答案: