Detecting Information Leakage in Predictive Process Mining: A Three-Layer Evaluation Framework
Abstract
Predictive process monitoring aims to provide early warnings about process outcomes, yet published benchmarks frequently report near-perfect prediction scores on full process traces. This paper investigates whether those scores reflect genuine predictive ability or information already embedded in the event log. A simulated supply-chain dataset with 818,292 events across 25,000 linked cases was used to evaluate 276 model configurations under controlled conditions where ground truth causal labels were available. Three forms of information leakage were identified- activity names that encode outcomes, conformance features that overlap with prediction targets, and temporal variables that measure downstream effects rather than upstream causes. When these sources were retained, models achieved an AUC of 1.00. After leakage was controlled through prefix-based evaluation, scores fell to between 0.76 and 0.93. A diagnostic procedure with specific detection thresholds is proposed for each layer. The framework was examined on two external benchmark datasets from different domains, and all three leakage patterns were independently observed in both. These findings suggest that near-perfect classification scores in predictive process mining should be interpreted with caution, and that a systematic leakage audit should precede model comparison or deployment.
Keywords: Predictive Process Monitoring, Information Leakage Detection, Process Mining, Prefix-Based Evaluation, Event Log Analysis, Process Intelligence, Supply Chain Analytics
DOI: 10.54941/ahfe1008421
Cite this paper
More from this volume
- AI-Enhanced Live-Streamed English Language Learning: Designing and Evaluating an Intelligent Platform for Chinese Learners
- "They Will Move": Confirmation Bias, Expectation Frames, and Premature Monitoring Drop Off in COLREG Encounters
- Automation transparency communication does not improve supervisory control performance, but shapes operator trust
- The influence of neuroticism on startle and surprise response: a flight simulator study
- EEG-Based Classification of Reward, Penalty, and Movement Events Across Four Video Games of Varying Complexity
- The Effects of Task Similarity and Time Pressure on Task Performance Following Interruptions in Complex Human-Computer Interaction Systems
- Trusting the Chain, Not the Black Box: A Provenance-Based Architecture for Human-Machine Teaming in Military Autonomy
- Design of a Simplified Humanoid Robot Face: Cultural Differences and Emotional Synchronization
- From Visual Observation to Auditory Interaction: A Human-Centered Tangible Music System Based on Physical Objects
- The Impact of Scarcity-Based Dark Patterns on User Performance, Perceived Darkness, and Workload in Subscription Services
- A Behavioral Observation Tool for Diagnosing Human–AI Interaction Profiles in Engineering Design Education
- The Mediated Moon: Semantic Mediation Between Gesture and Generative Image


AHFE Open Access