---
title: "Leveraging Path Information to Generate Predictions for Parallel Business Processes"
type: source
tags: [ppm, decision-tree, semi-structured, parallel-paths, outcome]
authors: [Unuvar Merve; Lakshmanan Geetika T.; Doganata Yurdaer N.]
year: 2016
venue: "Knowledge and Information Systems 47: 433–461, Springer"
kind: paper
raw_path: "raw/Predictive process monitoring/Leveraging path information to generate predictions unuvar2015.pdf"
created: 2026-04-13
updated: 2026-04-13
key_claims:
  - For semi-structured processes with parallel paths, the sequence of executed tasks (the "path") materially influences case outcome.
  - Five different representations of execution-trace path attributes are compared.
  - Information-gain analysis determines when parallel paths can be treated as independent vs dependent.
  - Decision trees are used as the prediction model.
---

# Unuvar, Lakshmanan, Doganata 2016 — Path Information for Parallel Processes

Pre-deep-learning PPM targeting **semi-structured business processes with parallel execution paths**. Companion to [[sources/2013-lakshmanan-markov-semi-structured]] from the same IBM T.J. Watson research stream.

## Contribution
- Five models for representing the execution path as a feature for prediction.
- Information-gain-based test for independence of parallel paths.
- Decision-tree outcome prediction evaluated on a simulated marketing-campaign business process.

## Position
Part of the classical decision-tree lineage of PPM that preceded deep learning. Its insistence on **path representation** as a prediction feature resurfaces in modern suffix-prediction and [[concepts/trace-encoding|trace-encoding]] research.

## Connections
**Concepts:** [[concepts/outcome-prediction]] · [[concepts/trace-encoding]]
**Authors:** [[entities/merve-unuvar]] · [[entities/geetika-lakshmanan]]