In SQL Server the VAR() function returns the statistical variance of all values in the specified expression.
Syntax
The function can be used as an aggregate function or as an analytical/window function.
When used as an aggregate function, the syntax goes like this:
VAR ( [ ALL | DISTINCT ] expression )And as an analytical function:
VAR ([ ALL ] expression) OVER ( [ partition_by_clause ] order_by_clause)The expression must be either an exact numeric or approximate numeric data type, except for the bit data type.
Example
Suppose we have a table like this:
SELECT * FROM Dogs;Result:
DogName Activity Score ------- ----------- ----- Bruno Fetch Stick 43 Cooper Fetch Stick 43 Max Fetch Stick 91 Bruno Wag Tail 65 Cooper Wag Tail 51 Max Wag Tail 87 Bruno Keep Quiet 3 Cooper Keep Quiet null Max Keep Quiet 1 null Catch Cat null
We can use the VAR() function against the Score column like this:
SELECT VAR(Score) FROM Dogs;Result:
1136
The VAR() function ignores NULL values and so these aren’t factored into the results.
The DISTINCT Argument
By default, all rows are evaluated, regardless of whether they contain unique values. However, we can use the DISTINCT keyword to specify that only distinct (unique) values are evaluated:
SELECT VAR(DISTINCT Score) FROM Dogs;Result:
1320.5714285714287
We can see that this has changed the result slightly. This is because our underlying data contains duplicates – two rows contain 43. When we used the DISTINCT keyword, the VAR() function treated those two as one. This resulted in a different statistical variance.
The ALL Argument
By default, all rows are evaluated regardless of whether they contain duplicates. In other words, this is the case when we don’t use the DISTINCT argument to eliminate duplicates from evaluation.
However, we also have the option of using the ALL keyword to explicitly specify that all values are evaluated:
SELECT VAR(ALL Score) FROM Dogs;Result:
1136
We can see that this result is the same as the first example. That’s because the functionality is exactly the same when we use the ALL keyword.
The OVER Clause
We can include an OVER clause in order to use VAR() as a window function.
Example:
SELECT
Activity,
DogName,
Score,
VAR(Score) OVER(ORDER BY Activity) AS "VAR"
FROM Dogs;Result:
Activity DogName Score VAR ----------- ------- ----- ------ Catch Cat null null null Fetch Stick Bruno 43 768 Fetch Stick Cooper 43 768 Fetch Stick Max 91 768 Keep Quiet Bruno 3 1359.2 Keep Quiet Cooper null 1359.2 Keep Quiet Max 1 1359.2 Wag Tail Bruno 65 1136 Wag Tail Cooper 51 1136 Wag Tail Max 87 1136
Here, I used an ORDER BY clause in the OVER clause to specify the logical order in which the operation is performed.
We could also put another ORDER BY clause at the end of the query in order to determine how the result set is sorted.
Partitioning the Results
The OVER clause also accepts a PARTITION BY clause that allows us to partition the results by a given column.
Here’s an example:
SELECT
Activity,
DogName,
Score,
VAR(Score) OVER(PARTITION BY Activity) AS "VAR"
FROM Dogs;Result:
Activity DogName Score VAR ----------- ------- ----- ------------------ Catch Cat null null null Fetch Stick Bruno 43 768 Fetch Stick Cooper 43 768 Fetch Stick Max 91 768 Keep Quiet Bruno 3 2 Keep Quiet Cooper null 2 Keep Quiet Max 1 2 Wag Tail Bruno 65 329.33333333333303 Wag Tail Cooper 51 329.33333333333303 Wag Tail Max 87 329.33333333333303
In this case we partitioned by the Activity column, and so we got the statistical variance within each activity.
Here it is with an ORDER BY clause added to the OVER clause:
SELECT
Activity,
DogName,
Score,
VAR(Score) OVER(PARTITION BY Activity ORDER BY Score) AS "VAR"
FROM Dogs;Result:
Activity DogName Score VAR ----------- ------- ----- ------------------ Catch Cat null null null Fetch Stick Bruno 43 0 Fetch Stick Cooper 43 0 Fetch Stick Max 91 768 Keep Quiet Cooper null null Keep Quiet Max 1 null Keep Quiet Bruno 3 2 Wag Tail Cooper 51 null Wag Tail Bruno 65 98 Wag Tail Max 87 329.33333333333303
No DISTINCT Clause with the OVER Clause
We can’t use the DISTINCT clause when using the OVER clause.
Here’s what happens when we add the DISTINCT clause to the previous example:
SELECT
Activity,
DogName,
Score,
VAR(DISTINCT Score) OVER(PARTITION BY Activity ORDER BY Score) AS "VAR"
FROM Dogs;Result:
Msg 10759, Level 15, State 1, Line 5 Use of DISTINCT is not allowed with the OVER clause.
However, we can use the ALL clause if we want.