SHAining on Process Mining: Explaining Event Log Characteristics Impact on Algorithms

2025-09-10

AI Summary

Existing studies largely overlook the interdependencies among structural characteristics of event logs, hindering the isolation of their individual effects on process mining algorithm performance. Method: We propose SHAining—a marginal contribution analysis framework—that systematically quantifies the independent impact of key structural features (e.g., noise ratio, behavioral variability, completeness) on adaptability, precision, and model complexity across over 22,000 real and synthetic logs. Unlike conventional causal assumptions, SHAining explicitly models feature interaction effects, integrating statistical modeling with interpretable analysis in process discovery. Contribution/Results: SHAining reveals the magnitude and nonlinear patterns of feature influence, identifying the most impactful core features. It further enables robustness assessment of algorithms under varying structural conditions. The findings provide empirically grounded guidance for event log preprocessing, algorithm selection, and performance optimization—advancing both theoretical understanding and practical deployment of process mining techniques.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Causal Learning
• Knowledge Representation and Reasoning: Action, Change, and Causality

Application Category

• Web Mining and Content Analysis: Bridging structured and unstructured data
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics

Abstract

Process mining aims to extract and analyze insights from event logs, yet algorithm metric results vary widely depending on structural event log characteristics. Existing work often evaluates algorithms on a fixed set of real-world event logs but lacks a systematic analysis of how event log characteristics impact algorithms individually. Moreover, since event logs are generated from processes, where characteristics co-occur, we focus on associational rather than causal effects to assess how strong the overlapping individual characteristic affects evaluation metrics without assuming isolated causal effects, a factor often neglected by prior work. We introduce SHAining, the first approach to quantify the marginal contribution of varying event log characteristics to process mining algorithms' metrics. Using process discovery as a downstream task, we analyze over 22,000 event logs covering a wide span of characteristics to uncover which affect algorithms across metrics (e.g., fitness, precision, complexity) the most. Furthermore, we offer novel insights about how the value of event log characteristics correlates with their contributed impact, assessing the algorithm's robustness.

Problem

• Analyzing how event log characteristics impact process mining algorithms
• Quantifying marginal contributions of event log features to algorithm metrics
• Assessing algorithm robustness across diverse event log characteristics

Innovation

• SHAining quantifies event log characteristics' marginal contributions
• Analyzes 22000+ logs to identify key algorithm metric impacts
• Assesses algorithm robustness via characteristic value correlations