A comparative analysis of predictive process monitoring: object-centric versus classical event logs

Abstract Predictive Process Monitoring (PPM) techniques are emerging as part of the more general research scenario of Process Mining (PM). They play a crucial role in the continuously evolving process of digital transformation by constantly supporting the organizational decision-making processes providing (accurate) predictions on the future behavior of processes. The state of the art of PPM application methodologies is mainly focused on Single ID Event Logs, commonly known as Traditional Event Logs or Classical Event Logs. As a matter of fact, the importance of Object-Centric Event Logs (OCEL) is being increasingly recognized as many emerging PPM approaches benefited of the usage of OCEL by obtaining a significative increase of the prediction accuracy. This survey aims to explore the current proposals in the context of OCEL-based PPM approaches. More in detail, we contribute to the state of the art by adding new classification features by differentiating between the approaches based on the input Event Log (Traditional or OCEL). We also analyzed the existing literature considering the prediction task addressed, the methodology used, the specific contribution area they addressed and the application domain.