LAG Function [Analytical]

An inter-row function that returns the value of an attribute in a previous row in the table or table partition.

Syntax

LAG (value_expr) [, offset [, default]]) OVER ([PARTITION BY window partition] ORDER BY window ordering)

Parameters

Parameter

Description

value_expr

Table column or expression defining the offset data to return from the table.

offset

The number of rows above the current row, expressed as a non-negative exact numeric literal, or as a SQL variable with exact numeric data. The permitted range is 0 to 231.

default

The value to return if the offset value goes beyond the scope of the cardinality of the table or partition.

window partition

(Optional) One or more value expressions separated by commas indicating how you want to divide the set of result rows.

window ordering

Defines the expressions for sorting rows within window partitions, if specified, or within the result set if you did not specify a window partition.

Usage

The LAG function requires an OVER (ORDER_BY) window specification. The window partitioning clause in the OVER (ORDER_BY) clause is optional. The OVER (ORDER_BY) clause must not contain a window frame ROWS/RANGE specification.

You cannot define an analytic expression in value_expr. That is, you cannot nest analytic functions, but you can use other built-in function expressions for value_expr.

You must enter a non-negative numeric data type for offset. Entering 0 returns the current row. Entering a negative number generates an error.

The default value of default is NULL. The data type of default must be implicitly convertible to the data type of the value_expr value or else SAP Sybase IQ generates a conversion error.

Example

The following example returns salary data from the Employees table, partitions the data by department ID, and orders the data according to employee start date. The LAG function returns the salary from the previous row (a physical offset of one row) and displays it under the LAG (Salary) column:

SELECT DepartmentID dID, StartDate, Salary, LAG(Salary, 1) OVER(PARTITION BY dID ORDER BY StartDate) FROM Employees ORDER BY 1,2;

The returned result set is:

dID        StartDate    Salary      Lag(Salary)
=========  ===========  ==========  =============
100        1984-08-28   45,700.000  NULL
100        1985-01-01   62,000.000  45,700.000
100        1985-06-17   57,490.000  62,000.000
100        1986-06-07   72,995.000  57,490.000
100        1986-07-01   48,023.690  72,995.000
...
200        1985-02-03   38,500.000  NULL
200        1985-12-06   54,800.000  38,500.000
200        1987-02-19   39,300.000  54,800.000
200        1987-07-10   49,500.000  39,300.000
200        1988-10-04   54,600.000  49,500.000
200        1988-11-12   39,800.000  54,600.000 
...

Standards and Compatibility

  • SQL—Vendor extension to ISO/ANSI SQL grammar.

Related reference
LEAD Function [Analytical]