Towards a Business-Oriented Approach to Visualization-Supported Interpretability of Prediction Results in Process Mining

Abstract

The majority of the state-of-the-art predictive process monitoring approaches are based on machine learning techniques. However, many machine learning techniques do not inherently provide explanations to business process analysts to interpret the results of the predictions provided about the outcome of a process case and to understand the rationale behind such predictions. In this paper, we introduce a business-oriented approach to visually support the interpretability of the results in predictive process monitoring. We take as input the results produced by the SP-LIME interpreter and we project them onto a process model. The resulting enriched model shows which features contribute to what degree to the predicted result. We exemplify the proposed approach by visually interpreting the results of a classifier to predict the output of a claim management process, whose claims can be accepted or rejected.