How Will My Business Process Unfold? Predicting Case Suffixes with Start and End Timestamps

Predictive process monitoring techniques support operational decision-making by predicting future states of ongoing cases in a business process. A subset of these techniques focuses on predicting the remaining sequence of activities for an ongoing case, known as case suffix prediction. Existing approaches for case suffix prediction typically generate sequences of activities with a single timestamp (e.g., the end timestamp). While this approach is useful in some contexts, it is insufficient for applications like resource capacity planning, where it is critical to reason about both the waiting time and processing time of each activity. Such information is necessary to accurately assess when resources will be engaged, optimize scheduling, and ensure capacity is adequate for future workloads. This paper introduces a technique for predicting case suffixes consisting of activities with both start and end timestamps. Specifically, the proposed technique predicts both the waiting time and processing time of each activity. Since the waiting time of an activity in a case depends on the availability of resources in other ongoing cases, the technique adopts a sweep-line approach, wherein the suffixes of all ongoing cases are predicted in lockstep, rather than predictions being made for each case in isolation. Additionally, the technique improves temporal accuracy by integrating resource availability information, automatically discovered from historical event logs, allowing for more realistic predictions in dynamic environments. An evaluation on real-life and synthetic datasets compares the accuracy of different instantiations of this approach, demonstrating the advantages of a multi-model approach to case suffix prediction over existing methods. The results highlight the improved performance in predicting both control-flow and temporal aspects of case suffixes, particularly in scenarios involving sparse and intermittent resource availability.