The Partition analysis is one of two functions provided by Teradata Warehouse Miner to sample data from a table or view. The Partition analysis is distinguished from the Sample analysis in that it is repeatable and is based on the internal hash index encodings provided by Teradata, rather than the statistically random selections provided by the Sample function.
Given a table, a list of columns to select and a list of columns to hash on, the Partition analysis generates a user specific partition or range of partitions from a table using a hash key. For example, the 3rd partition out of 10 might be requested, or partitions 1 through 3 out of 10.
To select a specific partition, set start and end partition to the same selected value. If a range of partitions is requested, the partition number is also returned as xpartid.
The Partition analysis is parameterized by specifying the table and column(s) to analyze, options unique to the Partition analysis, as well as specifying the desired results and SQL or Expert Options.