Selecting Optimal Trace Clustering Pipelines with AutoML
This work addresses the need for automated pipeline selection in trace clustering to improve model understandability and conformance analytics, but it is incremental as it applies existing AutoML concepts to a specific domain.
The paper tackles the problem of selecting optimal trace clustering pipelines for event logs by proposing an AutoML framework that recommends encoding methods, clustering algorithms, and hyperparameters, with experiments on a thousand event logs showing it can assist users in choosing the best pipeline for their scenario.
Trace clustering has been extensively used to preprocess event logs. By grouping similar behavior, these techniques guide the identification of sub-logs, producing more understandable models and conformance analytics. Nevertheless, little attention has been posed to the relationship between event log properties and clustering quality. In this work, we propose an Automatic Machine Learning (AutoML) framework to recommend the most suitable pipeline for trace clustering given an event log, which encompasses the encoding method, clustering algorithm, and its hyperparameters. Our experiments were conducted using a thousand event logs, four encoding techniques, and three clustering methods. Results indicate that our framework sheds light on the trace clustering problem and can assist users in choosing the best pipeline considering their scenario.