Illuminating Process Predictions: A Visual Analytics Framework for Business Workflows

• Ujjwal Yadav

  Author

As organizations increasingly use machine learning in business process management, the need for transparent predictive models becomes critical. Predictive process monitoring (PPM) techniques deliver accurate forecasts of process outcomes, remaining time and compliance violations. However, the black-box nature of many algorithms hinders their real-world deployment. Decision-makers in regulated industries want not only accurate predictions, but also understandable explanations for those predictions. VisInterPPM is a visual analytics framework introduced in this paper to help with the interpretability of predictive process monitoring. The structure involves predictive modeling through Random Forest classifiers, post-hoc explanation creation with LIME software, and visual-to-model mapping on Business Process Model and Notation diagrams. VisInterPPM allows analysts to comprehend which activities are responsible for predictions on a per-case and cohort level by mapping local feature influence weights onto familiar representations of the process. This framework allows analysis at the global and local levels. A thorough assessment with both synthetic and real-life event logs confirms the framework's performance. Professional assessments reveal greater understandability, usefulness, ease of use, speed of insight, and confidence in VisInterPPM visual explanations compared to standard bar charts. The findings confirm that adding visual context to the process flow significantly aids business stakeholders in understanding. This research proposes a human-centred practical approach to making predictive process monitoring more accessible, actionable and trusted.

• PDF

Copyright (c) 2026 International Journal of Adaptive Management and Business Intelligence

This work is licensed under a Creative Commons Attribution 4.0 International License.