Go Premium for a chance to win a PS4. Enter to Win

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 140
  • Last Modified:

Help with Max function

Hi..
Trying to update a column with the max value of 3 other columns
what is wrong with this?

UPDATE  SALES  SET HIGHEST_OF_3  
 = (SELECT MAX(v) FROM VALUES (JON), (JANE), (BOB)) AS VALUE(v))

Incorrect syntax near the keyword 'values'.


thx
0
JElster
Asked:
JElster
2 Solutions
 
FloraCommented:
(SELECT Max(v)
   FROM (VALUES (JON), (JANE), (BOB)) AS value(v)) as Maxvalue
0
 
ste5anSenior DeveloperCommented:
The syntax of the Table Value Constructor is slightly different:

UPDATE  SALES
SET     HIGHEST_OF_3 = ( SELECT MAX(v)
                         FROM   ( VALUES ( 'JON' ), ( 'JANE' ), ( 'BOB' ) ) AS V ( v )
                       );

Open in new window


There were also the quotation marks missing.
0
 
PortletPaulCommented:
SQL Server does not have a "greatest()" function which works across the values of a row. The following is for SQL Server 2005. Later versions have other capabilities.

You could use a case expression to compare each column, but if you have more than 3 then this is going to get tedious very quickly.

I suggest the following:
update sales 
set [HIGHEST_OF_3] = m.mp
from sales
join (
      select id, max(person) mp
      from (
            select id, [JON] as person
            from sales
            union all
            select id, [JANE] as person
            from sales
            union all
            select id, [BOB] as person
            from sales
            ) as p
      group by id
    ) as m on sales.id = m.id
;

Open in new window

See a small sample running here: http://sqlfiddle.com/#!3/124d7/4

full setup:
**MS SQL Server 2008 Schema Setup**:

    
    
    CREATE TABLE SALES
    	([id] int identity primary key,  [JON] int, [JANE] int, [BOB] int, [HIGHEST_OF_3] int)
    ;
    	
    INSERT INTO SALES
    	([JON], [JANE], [BOB])
    VALUES
    	(11, 12, 13)
    ;

**Query 1**:

    update sales 
    set [HIGHEST_OF_3] = m.mp
    from sales
    join (
          select id, max(person) mp
          from (
                select id, [JON] as person
                from sales
                union all
                select id, [JANE] as person
                from sales
                union all
                select id, [BOB] as person
                from sales
                ) as p
          group by id
        ) as m on sales.id = m.id
    

**[Results][2]**:
    

**Query 2**:

    select
    *
    from sales
    
    

**[Results][3]**:
    
    | ID | JON | JANE | BOB | HIGHEST_OF_3 |
    |----|-----|------|-----|--------------|
    |  1 |  11 |   12 |  13 |           13 |



  [1]: http://sqlfiddle.com/#!3/124d7/4

  [2]: http://sqlfiddle.com/#!3/124d7/4/0

Open in new window

0
Industry Leaders: We Want Your Opinion!

We value your feedback.

Take our survey and automatically be enter to win anyone of the following:
Yeti Cooler, Amazon eGift Card, and Movie eGift Card!

 
ste5anSenior DeveloperCommented:
Using a GREATEST() function like PortletPaul demonstrated for T-SQL makes no sense from the relational view point. To compare those values they must be of the same kind, thus forming a repeating group.

Thus the my approach is here normally showing this:

CREATE TABLE SALES
    (
      id INT IDENTITY
             PRIMARY KEY ,
      JON INT ,
      JANE INT ,
      BOB INT
    );
    	
INSERT  INTO SALES
        ( JON, JANE, BOB )
VALUES  ( 11, 12, 13 );

INSERT  INTO SALES
        ( JON, JANE, BOB )
VALUES  ( 200, 300, 100 );

WITH    Normalized
          AS ( SELECT   *
               FROM     SALES S UNPIVOT ( Value FOR Attribute IN ( BOB, JANE, JON ) ) U
             ),
        MaxValue
          AS ( SELECT   N.id ,
                        MAX(N.Value) AS HIGHEST_OF_3
               FROM     Normalized N
               GROUP BY N.id
             )
    SELECT  S.* ,
            MV.HIGHEST_OF_3
    FROM    dbo.SALES S
            LEFT JOIN MaxValue MV ON MV.id = S.id;
   
DROP TABLE SALES;

Open in new window


This should be done in a view. When not other possible, then use a persisted computed column and use an explicit user defined function.
0
 
PortletPaulCommented:
There remains the issue of SQLServer 2005...
Values and pivot/unpivot dont exist in that version
0
 
ste5anSenior DeveloperCommented:
Sure?
0
 
PortletPaulCommented:
Im on a mobile phone right now, so its by memory. Pretty sure values () doesnt, and pivot/unpivot also.

Regardless, the 3 fiekds need to be placed in a column,  them MAX ()  can be used.
0
 
Scott PletcherSenior DBACommented:
If there are only 3 values, just add a computed column that uses "brute force" calc to determine it.  I assume JON can't be NULL and the others can, although the code should work regardless.

ALTER TABLE SALES
ADD HIGHEST_OF_3 AS CASE
    WHEN JANE > ISNULL(JON, '') AND JANE > ISNULL(BOB, '') THEN JANE
    WHEN BOB > ISNULL(JON, '') AND BOB > ISNULL(JANE, '') THEN BOB
    ELSE JON END
0

Featured Post

Keep up with what's happening at Experts Exchange!

Sign up to receive Decoded, a new monthly digest with product updates, feature release info, continuing education opportunities, and more.

Tackle projects and never again get stuck behind a technical roadblock.
Join Now