Implementing Analytics with Greenplum UAP - Getting Started with Greenplum for Big Data Analytics - page 145

A moving or rolling window defines a set of rows within a window partition. When you
define a window frame, the window function is computed with respect to the contents
of this moving frame, rather than against the fixed content of the entire window par-
tition. Window frames can be row-based, represented by the ROWS clause, or value-
based, represented by a RANGE clause.
When the window frame is row-based, you define the number of rows offset from the
current row. If the window frame is range-based, you define the bounds of the win-
dow frame in terms of data values offset from the value in the current row.
If you specify only a starting row for the window, the current row is used as the last
row in the window.
The OVER (ORDER BY…) clause
Greenplum supports a variety of methods for developing functions, including:
• Query language support for functions developed in SQL
• Procedural language support for functions written in languages such as PL/
pgSQL (which is a subset of PL/SQL), PL/TcL, Perl, Python, and R (a pro-
gramming)
• Language for statistical computing and graphics
• Internal functions
• C-language functions
Note
The data scientist may need to create a function that could be used in
the downstream analysis. Greenplum supports PL/pgSQL, PL/Perl, and
PL/Python out of the box. Other languages can be added with the cre-
atelang utility.
Creating, modifying, and dropping functions
Functions that operate on tables must be created in the same schema. If you modify
a table, we must have access to a schema.