Returns the sum of squares of the independent expressions used in a linear regression model. The REGR_SXX function can be used to evaluate the statistical validity of a regression model.
REGR_SXX( dependent-expression , independent-expression )
REGR_SXX( dependent-expression , independent-expression ) OVER ( window-spec )
window-spec : see Syntax 2 instructions in the Remarks section below
dependent-expression The variable that is affected by the independent variable.
independent-expression The variable that influences the outcome.
This function converts its arguments to DOUBLE, performs the computation in double-precision floating point, and returns a DOUBLE as the result. If the function is applied to an empty set, then it returns NULL.
The function is applied to the set of (dependent-expression and independent-expression) pairs after eliminating all pairs for which either dependent-expression or independent-expression is NULL. The function is computed simultaneously during a single pass through the data. After eliminating NULL values, the following computation is then made, where x represents the dependent-expression and y represents the independent-expression:
REGR_COUNT( x, y ) * VAR_POP( x )
For more information about the statistical computation performed, see Mathematical formulas for the aggregate functions.
Syntax 2 represents usage as a window function in a SELECT statement. As such, elements of window-spec can be specified either in the function syntax (inline), or in conjunction with a WINDOW clause in the SELECT statement. See the window-spec definition provided in WINDOW clause.
For more information about using window functions in SELECT statements, including working examples, see Window functions.
SQL/2003 SQL foundation feature (T621) outside of core SQL.
The following example returns the value 5916.4800000000105.
SELECT REGR_SXX( Salary, ( YEAR( NOW() ) - YEAR( BirthDate ) ) ) FROM Employees;
|Send feedback about this page via email or DocCommentXchange||Copyright © 2008, iAnywhere Solutions, Inc. - SQL Anywhere 11.0.0|