Scholarlink

<p>Machine learning (ML) interpretability is vital for advancing Predictive Process Monitoring (PPM) and making ML more actionable. The VisInter4PPM framework was designed to bridge the interpretability gap in PPM by visualizing insights on process predictions. This study evaluates VisInter4PPM’s effectiveness in offering interpretable predictions for process analysts. Evaluations involved experts in business process management (BPM), using interviews and self-assessment questionnaires, focusing on how insights from VisInter4PPM’s visualizations of process models enhance prediction interpretability. The evaluation showed that VisInter4PPM effectively meets business experts’ needs, improving interpretability through process models that clarify activity-level influences on predictions. This approach robustly supports decision-making in dynamic and multifaceted scenarios. The findings suggest ML’s role in BPM goes beyond automation, aiding human decision-making in complex settings. This study offers both a framework and empirical evidence to enhance ML transparency in BPM, thereby making it a key resource for practitioners guiding ML-driven process improvements.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Interpretability in Predictive Process Monitoring Using Process Models: An Expert Evaluation of the VisInter4PPM Framework

  • 期刊论文

  • 发表日期：

    2025年2月13日

查看全文
  • Ana Rocío Cárdenas Maita,
  • Marcelo Fantinato,
  • Sarajane Marques Peres,
  • Fabrizio Maria Maggi

摘要

Machine learning (ML) interpretability is vital for advancing Predictive Process Monitoring (PPM) and making ML more actionable. The VisInter4PPM framework was designed to bridge the interpretability gap in PPM by visualizing insights on process predictions. This study evaluates VisInter4PPM’s effectiveness in offering interpretable predictions for process analysts. Evaluations involved experts in business process management (BPM), using interviews and self-assessment questionnaires, focusing on how insights from VisInter4PPM’s visualizations of process models enhance prediction interpretability. The evaluation showed that VisInter4PPM effectively meets business experts’ needs, improving interpretability through process models that clarify activity-level influences on predictions. This approach robustly supports decision-making in dynamic and multifaceted scenarios. The findings suggest ML’s role in BPM goes beyond automation, aiding human decision-making in complex settings. This study offers both a framework and empirical evidence to enhance ML transparency in BPM, thereby making it a key resource for practitioners guiding ML-driven process improvements.