在 SQL 中查询此搜索的不同方法？答案

【问题标题】：Different ways to query this search in SQL?在 SQL 中查询此搜索的不同方法？
【发布时间】：2012-09-09 20:34:01
【问题描述】：

我正在自学 MS-SQL，我正在尝试从这 3 个表中找到不同的方法来查找按地区分组的 2012 年已付和未付索赔的计数。如果有返回日期，则未支付索赔，如果返回日期为空，则支付索赔。

我将附上我运行的代码，但我不确定是否有更好的方法。

谢谢。

代码如下：

SET dateformat ymd;

CREATE TABLE Claims
  (
     ClaimID      INT,
     SubID        INT,
     [Claim Date] DATETIME
  );

CREATE TABLE Phoneship
  (
     ClaimID           INT,
     [Shipping Number] INT,
     [Claim Date]      DATETIME,
     [Ship Date]       DATETIME,
     [Returned Date]   DATETIME
  );

CREATE TABLE Enrollment
  (
     SubID           INT,
     Enrollment_Date DATETIME,
     Channel         NVARCHAR(255),
     Region          NVARCHAR(255),
     Status          FLOAT,
     Drop_Date       DATETIME
  );

INSERT INTO [Phoneship]
            ([ClaimID],
             [Shipping Number],
             [Claim Date],
             [Ship Date],
             [Returned Date])
VALUES     (102,
            201,
            '2011-10-13 00:00:00',
            '2011-10-14 00:00:00',
            NULL);

INSERT INTO [Phoneship]
            ([ClaimID],
             [Shipping Number],
             [Claim Date],
             [Ship Date],
             [Returned Date])
VALUES     (103,
            202,
            '2011-11-02 00:00:00',
            '2011-11-03 00:00:00',
            '2011-11-20 00:00:00');

INSERT INTO [Phoneship]
            ([ClaimID],
             [Shipping Number],
             [Claim Date],
             [Ship Date],
             [Returned Date])
VALUES     (103,
            203,
            '2011-11-02 00:00:00',
            '2011-11-22 00:00:00',
            NULL);

INSERT INTO [Phoneship]
            ([ClaimID],
             [Shipping Number],
             [Claim Date],
             [Ship Date],
             [Returned Date])
VALUES     (105,
            204,
            '2012-01-16 00:00:00',
            '2012-01-17 00:00:00',
            NULL);

INSERT INTO [Phoneship]
            ([ClaimID],
             [Shipping Number],
             [Claim Date],
             [Ship Date],
             [Returned Date])
VALUES     (106,
            205,
            '2012-02-15 00:00:00',
            '2012-02-16 00:00:00',
            '2012-02-26 00:00:00');

INSERT INTO [Phoneship]
            ([ClaimID],
             [Shipping Number],
             [Claim Date],
             [Ship Date],
             [Returned Date])
VALUES     (106,
            206,
            '2012-02-15 00:00:00',
            '2012-02-27 00:00:00',
            '2012-03-06 00:00:00');

INSERT INTO [Phoneship]
            ([ClaimID],
             [Shipping Number],
             [Claim Date],
             [Ship Date],
             [Returned Date])
VALUES     (107,
            207,
            '2012-03-12 00:00:00',
            '2012-03-13 00:00:00',
            NULL);

INSERT INTO [Phoneship]
            ([ClaimID],
             [Shipping Number],
             [Claim Date],
             [Ship Date],
             [Returned Date])
VALUES     (108,
            208,
            '2012-05-11 00:00:00',
            '2012-05-12 00:00:00',
            NULL);

INSERT INTO [Phoneship]
            ([ClaimID],
             [Shipping Number],
             [Claim Date],
             [Ship Date],
             [Returned Date])
VALUES     (109,
            209,
            '2012-05-13 00:00:00',
            '2012-05-14 00:00:00',
            '2012-05-28 00:00:00');

INSERT INTO [Phoneship]
            ([ClaimID],
             [Shipping Number],
             [Claim Date],
             [Ship Date],
             [Returned Date])
VALUES     (109,
            210,
            '2012-05-13 00:00:00',
            '2012-05-30 00:00:00',
            NULL);

INSERT INTO [Claims]
            ([ClaimID],
             [SubID],
             [Claim Date])
VALUES     (101,
            12345678,
            '2011-03-06 00:00:00');

INSERT INTO [Claims]
            ([ClaimID],
             [SubID],
             [Claim Date])
VALUES     (102,
            12347190,
            '2011-10-13 00:00:00');

INSERT INTO [Claims]
            ([ClaimID],
             [SubID],
             [Claim Date])
VALUES     (103,
            12348723,
            '2011-11-02 00:00:00');

INSERT INTO [Claims]
            ([ClaimID],
             [SubID],
             [Claim Date])
VALUES     (104,
            12349745,
            '2011-11-09 00:00:00');

INSERT INTO [Claims]
            ([ClaimID],
             [SubID],
             [Claim Date])
VALUES     (105,
            12347190,
            '2012-01-16 00:00:00');

INSERT INTO [Claims]
            ([ClaimID],
             [SubID],
             [Claim Date])
VALUES     (106,
            12349234,
            '2012-02-15 00:00:00');

INSERT INTO [Claims]
            ([ClaimID],
             [SubID],
             [Claim Date])
VALUES     (107,
            12350767,
            '2012-03-12 00:00:00');

INSERT INTO [Claims]
            ([ClaimID],
             [SubID],
             [Claim Date])
VALUES     (108,
            12350256,
            '2012-05-11 00:00:00');

INSERT INTO [Claims]
            ([ClaimID],
             [SubID],
             [Claim Date])
VALUES     (109,
            12347701,
            '2012-05-13 00:00:00');

INSERT INTO [Claims]
            ([ClaimID],
             [SubID],
             [Claim Date])
VALUES     (110,
            12350256,
            '2012-05-15 00:00:00');

INSERT INTO [Claims]
            ([ClaimID],
             [SubID],
             [Claim Date])
VALUES     (111,
            12350767,
            '2012-06-30 00:00:00');

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12345678,
            '2011-01-05 00:00:00',
            'Retail',
            'Southeast',
            1,
            NULL);

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12346178,
            '2011-03-13 00:00:00',
            'Indirect Dealers',
            'West',
            1,
            NULL);

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12346679,
            '2011-05-19 00:00:00',
            'Indirect Dealers',
            'Southeast',
            0,
            '2012-03-15 00:00:00');

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12347190,
            '2011-07-25 00:00:00',
            'Retail',
            'Northeast',
            0,
            '2012-05-21 00:00:00');

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12347701,
            '2011-08-14 00:00:00',
            'Indirect Dealers',
            'West',
            1,
            NULL);

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12348212,
            '2011-09-30 00:00:00',
            'Retail',
            'West',
            1,
            NULL);

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12348723,
            '2011-10-20 00:00:00',
            'Retail',
            'Southeast',
            1,
            NULL);

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12349234,
            '2012-01-06 00:00:00',
            'Indirect Dealers',
            'West',
            0,
            '2012-02-14 00:00:00');

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12349745,
            '2012-01-26 00:00:00',
            'Retail',
            'Northeast',
            0,
            '2012-04-15 00:00:00');

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12350256,
            '2012-02-11 00:00:00',
            'Retail',
            'Southeast',
            1,
            NULL);

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12350767,
            '2012-03-02 00:00:00',
            'Indirect Dealers',
            'West',
            1,
            NULL);

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12351278,
            '2012-04-18 00:00:00',
            'Retail',
            'Midwest',
            1,
            NULL);

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12351789,
            '2012-05-08 00:00:00',
            'Indirect Dealers',
            'West',
            0,
            '2012-07-04 00:00:00');

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12352300,
            '2012-06-24 00:00:00',
            'Retail',
            'Midwest',
            1,
            NULL);

INSERT INTO [Enrollment]
            ([SubID],
             [Enrollment_Date],
             [Channel],
             [Region],
             [Status],
             [Drop_Date])
VALUES     (12352811,
            '2012-06-25 00:00:00',
            'Retail',
            'Southeast',
            1,
            NULL);

查询1

SELECT Count(ClaimID)                       AS 'Paid Claim',
       (SELECT Count(ClaimID)
        FROM   dbo.phoneship
        WHERE  [returned date] IS NOT NULL) AS 'Unpaid Claim'
FROM   dbo.Phoneship
WHERE  [Returned Date] IS NULL
GROUP  BY claimid

查询2

SELECT Count(*)                             AS 'Paid Claims',
       (SELECT Count(*)
        FROM   dbo.Phoneship
        WHERE  [Returned Date] IS NOT NULL) AS 'Unpaid Claims'
FROM   dbo.Phoneship
WHERE  [Returned Date] IS NULL;

查询3

Select Distinct(C.[Shipping Number]), Count(C.ClaimID) AS 'COUNT ClaimID', 
        A.Region, A.SubID 
From dbo.HSEnrollment A 
Inner Join dbo.Claims B On A.SubId = B.SubId 
Inner Join dbo.Phoneship C On B.ClaimID = C.ClaimID 
Where C.[Returned Date] IS NULL 
Group By A.Region, A.Subid, C.ClaimID, C.[Shipping Number] Order By A.Region

【问题讨论】：

您已包含用于创建数据库的脚本，但未包含用于获取所需值的脚本。您尝试查询什么？
用这个来编辑你的问题比将查询粘贴到 cmets 中更好（前两个我已经为你这样做了）
我知道这不能回答您的问题，但我想马上告诉您避免在列名中使用空格。它可能会导致麻烦。有些人使用下划线，但我更喜欢 Camel Case Notation en.wikipedia.org/wiki/CamelCase - 例如 [Shipping Number] 可以是 ShippingNumber，在编写查询时更容易键入，并且您的开发人员（如果有的话）不会想杀了你:D
对不起，我不想再竖琴了，只是另一个建议 - 不要使用 DateTime，使用 MS 推荐的 DateTime2：msdn.microsoft.com/en-us/library/ms187819.aspx 甚至在 StackOverFlow stackoverflow.com/questions/1334143/…
@dyslexicanaboko 为什么 CamelCasers 首字母大写？他们不应该像这样大写吗？ AndNotLikeThis？

标签： sql sql-server sql-server-2008 join

【解决方案1】：

这个问题很难回答，因为我明白你在问什么，但还有许多其他小问题会导致你的查询困难。

我的答案
因此，当且仅当我正确解释您的表结构时，我会在此处回答您问题的核心（更多内容将在此基础上进行）。

我没有包括索赔的数量，因为这会影响数字。我已经包含了两个查询，最终查询和分解查询。对于总和（因为您正在学习），我使用 WITH ROLLUP 来获取每个分组列的总和。

SELECT 
    e.Region,
    paid = SUM(CASE WHEN p.[Returned Date] IS NULL THEN 1 ELSE 0 END),
    unpaid = SUM(CASE WHEN p.[Returned Date] IS NULL THEN 0 ELSE 1 END)
FROM claims c
    INNER JOIN enrollment e
        ON e.SubID = c.SubID
    INNER JOIN phoneship p
        ON p.ClaimID = c.ClaimID
GROUP BY e.Region
WITH ROLLUP

崩溃
这是使用内部虚拟表（结果表-结果集等）的子选择的分解查询。我这样做是为了证明一个观点。

SELECT 
    x.Region,
    -- if the Returned Date is null then add 1, otherwise add 0
    paid = SUM(CASE WHEN x.ReturnedDate IS NULL THEN 1 ELSE 0 END),
    -- if the Returned Date is null then add 0, otherwise add 1
    unpaid = SUM(CASE WHEN x.ReturnedDate IS NULL THEN 0 ELSE 1 END)
FROM
(
-- Run this inner query to see what data you are actually working with
-- for your grouping/sums
SELECT 
    c.ClaimID,
    c.SubID,
    E = '#', -- This is just a separator
    e_SubID = e.SubID, -- This is equivalent to saying e.SubID AS e_SubID
    e.Region,
    P = '#', -- This is just a separator
    p_ClaimID = p.ClaimID,
    ShippingNo = p.[Shipping Number], -- Getting rid of those nasty spaces
    ReturnedDate = p.[Returned Date]
FROM claims c
    INNER JOIN enrollment e
        ON e.SubID = c.SubID
    -- Initially this was a LEFT JOIN but you are missing Claims in your 
    -- Phoneship table which will produce bogus results, therefore the
    -- INNER JOIN will filter out any rows that don't match
    INNER JOIN phoneship p
        ON p.ClaimID = c.ClaimID
) as x
GROUP BY x.Region
WITH ROLLUP

尽可能避免使用 Sub Selects。它们对性能不是很好，但当然也有避免不了的时候。

您的表结构/关系是您无法执行此查询的根本原因。在查看结构后，我发现您正在复制数据（这是不可以的）并且您无法将详细信息全部提取到 1 个不错的查询中。

我看到的问题领域（以及一些友好的建议）

您将 ClaimDate 列从 Claims 表复制到 PhoneShip 表。我不确定它们是否有不同的含义，但如果它是重复的 - 避免这种情况。
可能应该删除 Claims 表中的 SubID。如果您将 ClaimID 作为 Foreign Key (FK) 放入 Enrollment 表中会更好。
为 Phoneship 表提供自己的Primary Key (PK) - 它易于使用，除了 ClaimID 和 ShippingNumber 的组合之外，每一行都是唯一的。查看Table Relationships 和Unique Constraints。
对于使用 NULL 作为衡量某物是否已支付或未支付的良好指标，我有点怀疑。只有设计者会知道空字段意味着付费。为此，您最好使用默认值为零并标记为 NOT NULL 的位字段 - 这样它就永远不会为空。毕竟这将为您省去编写 case 语句的麻烦，您可以直接使用该位作为您的总和例如：SUM(x.Paid)。此外，由于各种原因，列有时会被错误地标记为 NULL。
考虑将 Channel 和 Region 列完全拉出 Enrollment 表。用整数 PK 将它们放在他们自己的表中。您可以使用 ChannelID 和 RegionID 在任何需要的地方引用 PK。这样，如果名称需要更改，您就不必担心数据完整性问题（UPDATE Table SET NameCol = 'a' WHERE NameCol = 'b' -- 这可能会导致意外的重命名灾难。）
将 RegionID 和 ChannelID 放入 Claims 表中。现在，如果您按照上面的步骤 2 操作，则在 Enrollment 表中不需要它（在 Enrollment 表中有一个 ClaimID FK）。

恭喜你主动学习了这些东西。这是无价的知识（除非你上大学，在这种情况下，它的价值约为 50K 或更差……学生贷款……叹息……）。

【讨论】：

【解决方案2】：

您需要将所有表连接在一起才能获得区域。此版本假定每个索赔最多有 Phoneship 记录：

SELECT e.region, count(*) as numclaims,
       sum(case when ps.ReturnedDate is not null then 1 else 0 end) AS 'Paid Claim',
       sum(case when ps.ReturnedDate is null then 1 else 0 end) AS 'Unpaid Claim'
FROM   claims c join
       enrollment e
       on c.sub_id = e.sub_id left outer join
       Phoneship ps
       on ps.claimid = c.claimdid
WHERE  [Returned Date] IS NULL
GROUP  BY e.region

如果有多个，则计数将被取消，因为将计算每个电话号码而不是每个索赔。要解决此问题，请将两个总和更改为：

count(distinct case when ps.ReturnedDate is not null then c.claimid end)
count(distinct case when ps.ReturnedDate is null then c.claimid end)

【讨论】：

【解决方案3】：

试试这个

SELECT e.Region, COUNT(c.SubID) TotalClaims, COUNT(p.[returned date]) UnpaidClaims, COUNT(c.SubID)-COUNT(p.[returned date]) PaidClaims
FROM
    Claims c
    INNER JOIN enrollment e ON c.SubID = e.SubID
    INNER JOIN phoneship p ON p.ClaimID = c.ClaimID
GROUP BY e.Region

【讨论】：