Question

早上好，

我正在努力将我们的仪表板团队的许多指标组合成一个可复制到Excel中的单个可执行存储过程，然后让数据透视表做一些魔术。话虽如此，为了这个问题，我只会引用一个较小的数据集来表示单个指标，以保持简单。他们想要标题和标签的方式是非常具体的，这就是为什么前几组以他们的方式编写的原因。如果需要，我可以提供更多信息。

以下是我想知道是否有人可以帮助我的问题：

他们想要查询中的放电月份名称。这很好，但是一旦我将放电日期（projected_d）添加到查询中，我还必须使用group by。然后，这会导致任何没有数据的Month或Measure从结果集中完全消失。我想为该线显示零而不是消失。如果我从查询中删除了month和group by，则会显示零。
由于ORDER BY，我无法对Months使用Union All语句。我想在结果集中显示月份名称，但按MONTH编号排序。在正常查询中，我通常不会遇到问题，但由于此查询中的Union All，我遇到了在何处或如何使其发挥作用的问题。

这是当前查询和结果集：

USE COMPANY_VHR_PROD
GO

DECLARE @MINDOS DATE,
        @MAXDOS DATE,
        @PRTFSuccess INT,
        @PRTFTotal INT

SET @MINDOS = '20170701'
SET @MAXDOS = '20170803'

SELECT @PRTFSuccess = COUNT(DISTINCT uniqueid_c)
FROM cd.enrollments
WHERE
    program_c = 'PRTF'
    AND level_c IN ('SIG', 'MOD')
    AND projected_d BETWEEN @MINDOS AND @MAXDOS

SELECT @PRTFTotal = COUNT(DISTINCT uniqueid_c)
FROM cd.enrollments
WHERE
    program_c = 'PRTF'
    AND projected_d BETWEEN @MINDOS AND @MAXDOS

SELECT
    '' AS "Program",
    '' AS "Measurement",
    '' AS "Month",
    '' AS "Count or Total",
    '' AS "Percentage or Average",
    '' AS "YTD"
WHERE 1=0

UNION ALL

SELECT
    '',
    '',
    '',
    '',
    '',
    ''

UNION ALL

SELECT
    'PRTF',
    'Client Response To Treatment',
    'Month',
    'Count',
    'Percentage',
    'YTD'

UNION ALL

SELECT
    'PRTF' AS "Program",
    CASE 
        WHEN e.level_c IS NULL THEN 'None Entered' 
        WHEN e.level_c = '' THEN 'None Entered' 
        WHEN e.level_c = 'SIG' THEN 'Significant Success Experienced'
        WHEN e.level_c = 'MOD' THEN 'Moderate Success Experienced'
        WHEN e.level_c = 'MILD' THEN 'Mild Success Experienced'
        WHEN e.level_c = 'INS' THEN 'Insignificant Success Experienced' 
        ELSE e.level_c END AS "Success",
    DATENAME(MONTH,e.projected_d),
    CONVERT(VARCHAR,COUNT(DISTINCT e.uniqueid_c),1) AS "Count",
    CONVERT(VARCHAR,CAST((COUNT(DISTINCT e.uniqueid_c) / CONVERT(DECIMAL(10,2),@PRTFTotal))*100 AS DECIMAL(10,2)),1),
    CONVERT(VARCHAR,SUM(COUNT(DISTINCT e.uniqueid_c)) OVER (PARTITION BY e.level_c),1)
FROM cd.enrollments e
WHERE
    e.program_c = 'PRTF'
    AND e.projected_d BETWEEN @MINDOS and @MAXDOS
GROUP BY e.level_c, DATENAME(MONTH,e.projected_d)

UNION ALL

SELECT
    'PRTF' AS "Program",
    'Total Clients Discharging' AS "Client Response to Treatment",
    'All',
    CONVERT(VARCHAR,COUNT(DISTINCT e.uniqueid_c),1) AS "Count",
    CONVERT(VARCHAR,CAST((COUNT(DISTINCT e.uniqueid_c) / CONVERT(DECIMAL(10,2),@PRTFTotal))*100 AS DECIMAL(10,2)),1),
    CONVERT(VARCHAR,COUNT(DISTINCT e.uniqueid_c),1) AS "YTD"
FROM cd.enrollments e
WHERE
    e.program_c = 'PRTF'
    AND e.projected_d BETWEEN @MINDOS and @MAXDOS

ResultSet

理想情况下，我希望它按此顺序显示这样的内容：

PRTF - Insignificant Success - July   - 0 -   0.00 - 0
PRTF - Insignificant Success - August - 0 -   0.00 - 0
PRTF - Mild Success          - July   - 0 -   0.00 - 0
PRTF - Mild Success          - August - 0 -   0.00 - 0
PRTF - Moderate Success      - July   - 1 -  25.00 - 1
PRTF - Moderate Success      - August - 0 -   0.00 - 1
PRTF - Significant Success   - July   - 2 -  50.00 - 3
PRTF - Significant Success   - August - 1 -  25.00 - 3
PRTF - Total Clients         - All    - 4 - 100.00 - 4

任何人都可以帮助我吗？

Answer 1

您可以使用子查询注入月份编号和排序顺序来处理标题，正文和页脚的排序。子查询仍将与UNION ALL组合，但它允许您只选择结果集的相关列，同时为排序提供使用元数据。

-- Load headers with sub-query to hide sort order attribute
SELECT
    [Program],[Measurement],[Month],[Count or Total],[Percentage or Average],[YTD]
FROM (
    SELECT '', '', '', '', '', '', 10, 0 -- SortOrder, MonthOrder
    UNION ALL SELECT 'PRTF', 'Client Response To Treatment', 'Month', 'Count', 'Percentage', 'YTD', 20, 0 -- SortOrder, MonthOrder
) Headers ([Program],[Measurement],[Month],[Count or Total],[Percentage or Average],[YTD], SortOrder, MonthOrder)

UNION ALL

-- Load records with sub-query to hide sort order attribute
SELECT
    [Program],[Measurement],[Month],[Count or Total],[Percentage or Average],[YTD]
FROM (
    SELECT
        'PRTF' AS "Program",
        CASE 
            WHEN e.level_c IS NULL THEN 'None Entered' 
            WHEN e.level_c = '' THEN 'None Entered' 
            WHEN e.level_c = 'SIG' THEN 'Significant Success Experienced'
            WHEN e.level_c = 'MOD' THEN 'Moderate Success Experienced'
            WHEN e.level_c = 'MILD' THEN 'Mild Success Experienced'
            WHEN e.level_c = 'INS' THEN 'Insignificant Success Experienced' 
            ELSE e.level_c END AS "Success",
        DATENAME(MONTH,e.projected_d),
        CONVERT(VARCHAR,COUNT(DISTINCT e.uniqueid_c),1) AS "Count",
        CONVERT(VARCHAR,CAST((COUNT(DISTINCT e.uniqueid_c) / CONVERT(DECIMAL(10,2),@PRTFTotal))*100 AS DECIMAL(10,2)),1),
        CONVERT(VARCHAR,SUM(COUNT(DISTINCT e.uniqueid_c)) OVER (PARTITION BY e.level_c),1),
        30, DATEPART(MONTH,e.projected_d) -- Add SortOrder, MonthOrder
    FROM cd.enrollments e
    WHERE
        e.program_c = 'PRTF'
        AND e.projected_d BETWEEN @MINDOS and @MAXDOS
    GROUP BY e.level_c, DATENAME(MONTH,e.projected_d)
    ) Records ([Program],[Measurement],[Month],[Count or Total],[Percentage or Average],[YTD], SortOrder, MonthOrder)

UNION ALL

-- Load footer with sub-query to hide sort order attribute
SELECT
    [Program],[Measurement],[Month],[Count or Total],[Percentage or Average],[YTD]
FROM (
    SELECT
        'PRTF' AS "Program",
        'Total Clients Discharging' AS "Client Response to Treatment",
        'All',
        CONVERT(VARCHAR,COUNT(DISTINCT e.uniqueid_c),1) AS "Count",
        CONVERT(VARCHAR,CAST((COUNT(DISTINCT e.uniqueid_c) / CONVERT(DECIMAL(10,2),@PRTFTotal))*100 AS DECIMAL(10,2)),1),
        CONVERT(VARCHAR,COUNT(DISTINCT e.uniqueid_c),1) AS "YTD",
        40, 0 -- Add SortOrder and month order
    FROM cd.enrollments e
    WHERE
        e.program_c = 'PRTF'
        AND e.projected_d BETWEEN @MINDOS and @MAXDOS
    ) Footers ([Program],[Measurement],[Month],[Count or Total],[Percentage or Average],[YTD], SortOrder, MonthOrder)

ORDER BY
    SortOrder, [Program], [Measurement], [MonthOrder]

多个联盟声明中的分区和排序

1 个答案: