The data Reorganization functions provide the ability to join, merge and denormalize preprocessed results into a wide analytic data set, as well as select a subset of the rows in a table. The result of these functions is a new restructured table that has been built from one or more existing tables, and/or a subset of the rows in a table.
The Sampling and Partitioning functions build a new table containing randomly selected rows in an existing table or view. Sampling is useful when it becomes unwieldy to perform an analytic process because of the high volume of data available. This is especially true for compute intensive analytic modeling tasks. Partitioning is similar to sampling but allows mutually distinct but all inclusive subsets of data to be requested by separate processes.
In the case of the Data Reorganization functions, NULL values are passed back as NULL. A special case is the Denorm Analysis which allows you to convert NULL values to zero.
- Denorm — Create new table denormalizing by removing key column(s).
- Join — Join tables or views by columns into a combined result table.
- Merge — Merge tables or views by rows into a combined result table.
- Partition — Select partition(s) from a table using a hash key.
- Sample — Select sample(s) from a table by size or fraction.
In order to add a Data Reorganization Analysis to a Teradata Warehouse Miner Data Mining Project, create a new analysis as described in Adding Analyses to a Project in the Teradata Warehouse Miner User Guide (Volume 1), B035-2300.
Select Reorganization from the menu.