Question

我试图从三个表中获取访客的年龄：按客户ID包含viist日期的访问，以及拥有客户端ID的客户端，以及拥有诊所ID的诊所。

这是我的sql语句：

DECLARE @Clinicname nvarchar(50) 
SET @Clinicname='First Clinic'

SELECT CASE 
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 10 THEN '1-10' 
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 20 THEN '11-20' 
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 30 THEN '21-30'
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 40 THEN '31-40' 
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 50 THEN '41-50'
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 60 THEN '51-60'
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 70 THEN '61-70'
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 80 THEN '71-80'
         ELSE '81+' 
       END AS age, 
       COUNT(*) AS n

FROM  Visit v 
          INNER JOIN  Client c ON v.ClientID = c.ClientID
          INNER JOIN  Clinic r ON v.ClinicId = r.ClinicId
          WHERE r.Name IN (@Clinicname)   
GROUP BY CASE 
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 10 THEN '1-10' 
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 20 THEN '11-20' 
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 30 THEN '21-30'
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 40 THEN '31-40' 
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 50 THEN '41-50'
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 60 THEN '51-60'
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 70 THEN '61-70'
         WHEN DATEDIFF(YEAR, c.BirthDate, MIN(v.Date)) <= 80 THEN '71-80'
         ELSE '81+'
         END

这是错误：不能在GROUP BY子句列表的组表达式中使用聚合或子查询。

当我用GETDATE（）替换MIN（v.Date）时，它可以工作。

Answer 1

一些猜测，基于松散的规格，未知版本等。一条建议：DATEDIFF(YEAR不是计算年龄的可靠方法。想想这个人的生日是在12月并且他们在1月访问的情况，反之亦然。 DATEDIFF只计算已经越过了多少边界，并不关心这个人的生日是否已经发生。

您可能可能会将这些查询片段中的一些组合在一起，但将它们分块可能会使逻辑更容易消化。

SQL Server 2005 +

;WITH vis AS
(
    SELECT 
        v.ClientId, 
        FirstVisit = MIN(v.[Date]), 
        NumVisits = COUNT(*)
    FROM dbo.Visit AS v
    INNER JOIN dbo.Clinic AS c
    ON c.ClinicId = v.ClinicId
    WHERE c.Name IN (@Clinicname)
    GROUP BY v.ClientId
),
ages AS (
    SELECT 
        v.ClientId,
        rough_age = DATEDIFF(YEAR, c.BirthDate, v.FirstVisit),
        v.NumVisits
    FROM vis AS v
    INNER JOIN dbo.Client AS c
    ON v.ClientId = c.ClientId
),
cats([group], numVisits) AS (
    SELECT 
        CASE WHEN rough_age/10 > 8 THEN '81+' ELSE
        CONVERT(VARCHAR(32), ((rough_age/10)+1)*10-9) + '-'
        + CONVERT(VARCHAR(12), ((rough_age/10)+1)*10) END,
        numVisits
   FROM ages
)
SELECT [group], NumClients = COUNT(*), NumVisits = SUM(numVisits)
FROM cats
GROUP BY [group];

SQL Server 2000

SELECT [group], NumClients = COUNT(*), NumVisits = SUM(numVisits)
FROM (
    SELECT 
        [group] = CASE WHEN rough_age/10 > 8 THEN '81+' ELSE
        CONVERT(VARCHAR(32), ((rough_age/10)+1)*10-9) + '-'
        + CONVERT(VARCHAR(12), ((rough_age/10)+1)*10) END,
        numVisits
   FROM 
   (
    SELECT 
        v.ClientId,
        rough_age = DATEDIFF(YEAR, c.BirthDate, v.FirstVisit),
        v.NumVisits
    FROM 
    (
        SELECT 
            v.ClientId, 
            FirstVisit = MIN(v.[Date]), 
            NumVisits = COUNT(*)
        FROM dbo.Visit AS v
        INNER JOIN dbo.Clinic AS c
        ON c.ClinicId = v.ClinicId
        WHERE c.Name = @Clinicname
        GROUP BY v.ClientId
    ) AS v
    INNER JOIN dbo.Client AS c
    ON v.ClientId = c.ClientId
  ) AS ages
) AS cats
GROUP BY [group];

Answer 2

您可以使用内部查询解决分组问题。在任何情况下，这都是首选，因为您不必复制存储逻辑。

我也删除了MIN（v.Date），因为我认为这不是必要的。

DECLARE 
     @Clinicname nvarchar(50);

SET @Clinicname='First Clinic'


select age, count(*) from
(
    SELECT CASE 
         WHEN DATEDIFF(YEAR, c.BirthDate, v.Date) <= 10 THEN '1-10' 
         WHEN DATEDIFF(YEAR, c.BirthDate, v.Date) <= 20 THEN '11-20' 
         WHEN DATEDIFF(YEAR, c.BirthDate, v.Date) <= 30 THEN '21-30'
         WHEN DATEDIFF(YEAR, c.BirthDate, v.Date) <= 40 THEN '31-40' 
         WHEN DATEDIFF(YEAR, c.BirthDate, v.Date) <= 50 THEN '41-50'
         WHEN DATEDIFF(YEAR, c.BirthDate, v.Date) <= 60 THEN '51-60'
         WHEN DATEDIFF(YEAR, c.BirthDate, v.Date) <= 70 THEN '61-70'
         WHEN DATEDIFF(YEAR, c.BirthDate, v.Date) <= 80 THEN '71-80'
         ELSE '81+' 
       END AS age
    FROM  Visit v 
          INNER JOIN  Client c ON v.ClientID = c.ClientID
          INNER JOIN  Clinic r ON v.ClinicId = r.ClinicId
          WHERE r.Name IN (@Clinicname) 
) t group by age;

SQL GROUP BY CASE

2 个答案: