DyLoPro: Profiling the Dynamics of Event Logs.

Modern business processes are often characterized by continuous change, which can lead to bias in the results of process mining techniques that assume a static process. This bias is caused by concept drift, which can manifest in many forms and affect various process perspectives. Current research on concept drift in process mining has focused on drift detection techniques in the control-flow perspective, with limited capabilities for comprehensive dynamic profiling of event logs. To address this gap, this paper presents the DyLoPro framework, a generic approach that facilitates the exploration of event log dynamics over time using visual analytics . The framework caters to all types of event logs and allows for the exploration of event log dynamics from various process perspectives, both individually and combined with the performance perspective. Additionally, the framework is accompanied by an efficient and user-friendly Python library, rendering it a valuable instrument for both researchers and practitioners. A case study using large real-life event logs demonstrates the effectiveness of the framework.