Skip to content
PlaidCloud

Table KNIME Rule Engine

Description

Section titled “Description”

The Table: KNIME Rule Engine step applies a list of KNIME Rule Engine rules that are held in a table, read when the step runs. The first rule that matches a row decides its outcome. Migrated KNIME workflows use this step for the Rule Engine (Dictionary), Rule Engine Variable (Dictionary), Rule-based Row Filter (Dictionary), and Rule-based Row Splitter (Dictionary) nodes, including PMML rule sets, when the rules come from a table that is only known when the workflow runs.

The step converts the rules it reads the same way the converter does. A rule it cannot carry, such as one that uses Java-only syntax or a flow variable, fails the step rather than writing a wrong column.

The step form has a Tables section where Use the Rules To picks the rule kind and you choose the source, output and rules tables, an Output section, a Rule Set Scoring section for PMML rule sets, and a Columns section.

Configuration

Section titled “Configuration”
  • source and target_frame — the table to read and the table to write. source_columns and target_columns are the columns read and written; the target column named by output_column gets the rules’ result.
  • rules_source — the table whose rows are the rules. rules_column holds each rule, outcome_column its outcome, and rules_order_column, when set, orders the rules.
  • rule_kind — engine (default) writes the rules’ outcome to output_column, typed by output_dtype. filter and splitter write the rows the rules keep: include (default true) decides whether a rule outcome of TRUE keeps or drops the row, and port is 1 for the rows kept or 2 for the rest. variable writes the one-row table of the value the rules compute, with no source table.
  • outcome_cell — the KNIME cell type of the outcome column, which decides how an outcome is written in a rule. knime_types gives the KNIME cell type of zoned date-time, duration and period input columns.
  • dollar_references — outcomes in $column$ form name columns rather than text. disallow_long (default true) matches KNIME’s setting of the same name.
  • PMML rule sets — default_rule is a rule that always fires, appended last. weight_column and default_weight (default 1.0) set each rule’s weight for weighted selection, where the heaviest rule is tried first. confidence_column, confidence_cell, confidence_output and default_confidence_rule write the confidence of each rule’s result.
Section titled “Related”