Predictive Business Process Monitoring with Transfer Learning

Monitoring business processes is crucial for detecting inefficiencies, bottlenecks, or deviations from the expected process flow. Predictive Business Process Monitoring is a field within process mining that focuses on forecasting future events and outcomes based on historical and real-time process data. On the other hand, Transfer Learning enables a model trained on one task to effectively apply its knowledge to a different, but related, task. Despite Transfer Learning’s potential in the context of predictive business process monitoring, it is still an underexplored area. This work contributes to bridging Predictive Business Process Monitoring and Transfer Learning. Unlike existing related studies that require full retraining for every new dataset, this study evaluates and compares how multiple ML and DL models behave under transfer learning conditions by demonstrating how knowledge acquired from one event log can be effectively transferred to structurally similar but behaviorally diverse processes. In this way, it provides a systematic, multi-log empirical assessment of Transfer Learning effectiveness in Predictive Business Process Monitoring.