Question

In MySQL I used to be able to do something like:

SELECT Something1.ID, Something2.AverageValue FROM
(SELECT ID, Max(Date) FROM Table GROUP BY ID) Something1
LEFT JOIN
(SELECT ID, AverageValue FROM Table) Something2
ON Something1.ID = Something2.ID

So what I am trying to do is give me the most recent averagevalue, per ID

The syntax in SQL Server appears to be different and doesn’t allow me to do the nested SELECT statements?

Answer 1

WITH Data AS
(
 SELECT ROW_NUMBER() OVER (PARTITION BY ID ORDER BY Date DESC) AS Row_Number,
 ID,
 AverageValue
 FROM Table
)
SELECT * FROM Data WHERE Row_Number = 1

This is the way you should do it in SQL Server. It makes “last/first/max/min/greatest-n-per group” easy as pie.

PARTITION BY keyword here equals your GROUP BY statement, and the ORDER BY here would translate into the MAX aggregate function you used in MySQL

Using this would enable you to do some fancy stuff like:

WITH Data AS
    (
     SELECT ROW_NUMBER() OVER (PARTITION BY ID ORDER BY Date DESC) AS Row_Number,
     ID,
     AverageValue
     FROM Table
    )

SELECT Newest.ID, 
Newest.AverageValue, 
(Newest.AverageValue - Next_Newest.AverageValue) AS AverageValue_Change,
Newest.Date
 FROM Data Newest
LEFT OUTER JOIN Data Next_Newest 
 ON Newest.Row_Number+1 = Next_Newest.Row_Number
AND Newest.ID = Next_Newest.ID 

WHERE Newest.Row_Number=1