---
title: Lasagna vs Spaghetti Processes
type: concept
tags: [bpm, process-mining, process-regularity, log-characterisation, taxonomy]
sources: ["[[sources/2011-vanderaalst-process-mining-book]]"]
created: 2026-04-20
updated: 2026-04-20
---

# Lasagna vs Spaghetti Processes

A widely-used process-mining shorthand for **process regularity**, introduced in [[sources/2011-vanderaalst-process-mining-book]] (Chs. 13–14). The metaphor captures the visual impression of the discovered process model.

## Lasagna processes (Ch. 13)

**Layered, well-structured** processes with limited variability:
- Most cases follow a small number of common paths.
- Clearly identifiable phases / layers.
- Discovered models are readable by domain experts.
- Typical examples: insurance claims, loan applications, procurement, regulated healthcare pathways.
- All standard mining techniques (discovery, conformance, performance) work well.

Van der Aalst proposes a **5-stage methodology** (§13.3) for lasagna processes: plan/justify → extract → control-flow model + connect log → integrated multi-perspective model → operational support.

## Spaghetti processes (Ch. 14)

**Unstructured, high-variability** processes where the discovered model looks like a tangled plate of spaghetti:
- Hundreds of distinct variants, long tails of rare paths.
- Typical of knowledge work, ad-hoc coordination, emergency response, debugging workflows, patient treatment in teaching hospitals.
- Direct discovery produces illegible models; **fuzzy mining** (Günther & van der Aalst) and **filtering / abstraction** are essential.
- Conformance and prediction are harder; often require variant clustering or declarative (rather than imperative) modelling — see [[frameworks/declare]].

## Why the dichotomy matters

The lasagna/spaghetti distinction is the **primary practical filter** applied to any event log before choosing a mining strategy:
- **Lasagna** → imperative discovery (inductive miner / BPMN) + alignment-based conformance + performance annotation.
- **Spaghetti** → fuzzy miner / filtering / clustering / declarative mining.

It also maps onto the [[concepts/behavioral-variability|behavioural-variability]] axis: highly spaghetti processes are often candidates for [[concepts/agentic-bpm|agentic]] automation where [[concepts/framed-autonomy|framed autonomy]] copes with variability the process designer cannot enumerate.

## Related
[[methods/process-mining-basics]] · [[concepts/process-discovery]] · [[concepts/behavioral-variability]] · [[concepts/process-model-quality]] · [[frameworks/declare]] (preferred for spaghetti)