Analyzing Delay Detection Difficulty in Business Processes

An arXiv paper (2608.14367) recently explores the inherent challenges in identifying delays within business processes, a vital skill for organizations to prevent missed deadlines and service level breaches. Analyzing 14 event logs, the research indicates that remaining times are often significantly right-skewed, with only a minor percentage of instances showing substantial delays. The authors contend that despite advancements in predictive process monitoring (PPM) through advanced deep learning techniques, aggregate metrics fail to provide comprehensive insights into model efficacy across the target distribution, particularly for critical cases. This study addresses these issues, enhancing the understanding of delay detection obstacles and potentially informing future enhancements in PPM systems. The findings are available on arXiv under the identifier 2608.14367.