Automatically reconciling the trade-off between prediction accuracy and earliness in prescriptive business process monitoring

Prescriptive business process monitoring provides decision support to managers on when and how adapt an ongoing prevent or mitigate undesired outcome. We focus the problem of automatically reconciling trade-off between prediction accuracy earliness in determining adapt. Adaptations should happen sufficiently early provide enough lead time for adaptation become effective. However, earlier predictions are typically less accurate than later predictions. This means that acting may unnecessary adaptations missed adaptations. Different approaches were presented literature reconcile earliness. So far, these compared with different baselines, evaluated using data sets even confidential sets. limits comparability replicability makes it difficult choose a concrete approach practice. perform comparative evaluation main alternative Using four public real-world event log two types models, we assess compare cost savings approaches. The experimental results indicate which criteria affect effectiveness help us state initial recommendations selection