PFA Predictor

This node applies a scoring engine expressed in PFA format to a KNIME data table.

For more information please visit the DMG PFA website. Supported primitive types are: boolean, int, long, float, double, string. Supported complex types are: Records, Maps, Union (Union types are used for nullable types in PFA. Other use cases are currently not supported).

Maps and records are mapped to table rows, so that the column name is the key and the cell in each row corresponds to a value. Arrays are mapped to collection cells.

If the scoring engine outputs scalar values, the output is a table with a single column with the name set in the "Output Column Name" option. If the output is a record or a map, then there is no need to set the column name, since it will be automatically set according to the column names from the PFA document.

Please note that this node does not output the original table with the PFA scoring engine's output appended, but just the PFA output.