Remaining Time Prediction for Processes with Inter-case Dynamics

Abstract Process mining techniques use event data to describe business processes, where the provided insights are used for predicting processes’ future states ( Predictive Monitoring ). Remaining Time Prediction of process instances is an important task in field (PPM). Existing approaches have two key limitations developing Models (RTM): (1) The features predictions lack context, and created models black-boxes. (2) considered be isolation, despite fact that states, e.g., number running instances, influence remaining time a single instance. Recent improve quality RTMs by utilizing context related batching-at-end inter-case dynamics process, using batching as feature. We propose approach decreases previous approaches’ reliance on user knowledge discovering fine-grained behavior. Furthermore, we enrich our with extracted multiple performance patterns (caused dynamics), which increases interpretability models. assess proposed prediction method real-world logs. Incorporating into results more accurate interpretable predictions.