Causal Graph Miner

This node implements the first step of the Hybrid Miner to discover a causal graph from an event table. A causal graph consists of nodes representing activities and two types of directed edges connecting nodes. Certain edges represent strong causal dependencies and uncertain edges represent weak dependencies. A third type of edges is used to represent long-term dependencies.

Options

Trace classifer

The column to be used as a trace classifier.

Event classifer

The column to be used as an event classifier.

Minimal activity frequency

An activity will be included if it occurs in at least x% of cases; set to 0 to include all activities.

Minimal trace variant frequency

A trace variant will be included if it covers at least x% of cases; set to 0 to include all trace variants.

Strong causality threshold

Lower bound for a strong causality between two activities.

Weak causality threshold

Lower bound for a weak causality between two activities; >set to 100% to avoid uncertain edges.

Long-term dependency threshold

Lower bound for a strong long-term causality between two activities.

Causality weight threshold

High values mean more emphasis on the split and join behavior of activities; low values mean more emphasis on the detection of concurrency and loops.

Views

Interactive View: Causal Graph

Workflows

• 00_Process_Mining_with_Community_NodesKNIME Hub
• Discovery of Hybrid Process ModelsKNIME Hub

Links

• No links available

Developers