Window functions apply to a subset of rows (a window) relative to the current row for computational purposes. These functions can be applied to the following transforms:
When window functions are used, each row in the windowed output should be identified by a primary key. This primary key, which identifies unique rows, can be specified in one of the following ways:
Order byfield indicates unique row identifiers. This field can contain one or more column references.
Group byfields can create a unique identifier for each row.
This primary key requirement applies to window functions used with any supported transform.
NOTE: Null values are ignored as inputs to these functions.