Conformance checking
Conformance checking is the formal process-mining technique of comparing an event log to a reference process model and producing fitness, precision, generalisation, and simplicity scores. The output quantifies how well the model represents what's actually happening — and where the divergences concentrate.
Conformance is one of the four core process mining activities (discovery, conformance, enhancement, operational support). The fitness score (0-1) measures how much of the log the model can replay; low fitness means the model misses real behaviour. The precision score measures how much the model allows that the log never shows; low precision means the model is too permissive. The trade-off matters because models that score 1.0 on both for any non-trivial log are usually overfit. The pragmatic target: fitness above 0.85, precision above 0.7, and explanations for the major non-fitting variants.
Related terms
- Process conformance
Process conformance is the analysis of how well an observed event log matches a reference process model — measuring deviations, missing steps, and extra steps.
- Process discovery
Process discovery is the process-mining technique of constructing a process model (typically BPMN or a Petri net) from an event log without prior knowledge of the intended process.
- Deviation analysis
Deviation analysis quantifies how often and how much an observed process diverges from its reference model — measuring missed steps, extra steps, out-of-order steps, and loops.