Defining and visualizing process execution variants from partially ordered event data
Daniel Schuster, Francesca Zerbato, Sebastiaan J. van Zelst, Wil M. P. van der Aalst
Abstract
The execution of operational processes generates event data stored in enterprise information systems. Process mining techniques analyze such event data to obtain insights vital for decision-makers to improve the reviewed process. In this context, event data visualizations are essential. We focus on visualizing variants describing process executions that are control flow equivalent. Such variants are an integral concept for process mining and are used, e.g., for data exploration and filtering. We propose high-level and low-level variants covering different levels of abstraction and present corresponding visualizations. Compared to existing variant visualizations, we support partially ordered event data and allow for heterogeneous temporal information per event, i.e., we support both time intervals and time points. We evaluate our contributions using automated experiments showing practical applicability to real-life event data. Finally, we present a user study indicating significantly improved usefulness and ease of use of the proposed high-level variant visualization compared to existing variant visualizations for typical analysis tasks.