In the design of a custom transformer, which attribute is cited as necessary to create parallel processing groups?

• A. Postcode attribute
• B. Input ID
• C. Area code
• D. Geometry

Answer: (A)

Parallel processing groups in a custom transformer are built by using a grouping attribute that partitions features into independent streams, so multiple groups can be processed at the same time. The postcode attribute serves as a good grouping key because it typically divides data into geographic regions with many features sharing the same code. This creates several distinct groups that can run in parallel, while still keeping features within the same group together for any group-internal operations.

Using a value that is unique for every feature, like an input ID, would effectively create a different group for each feature and eliminate the benefit of parallelism. Area code could work as a grouping key in the same way as postcode, but it’s functionally similar to using a region-based field. Geometry isn’t a straightforward attribute to group by, and basing groups on geometric values isn’t practical or stable for consistent parallel processing.

So, the postcode attribute is the best fit because it provides a stable, region-based grouping that enables meaningful parallel processing without collapsing all features into individual groups.