Predictive Process

Predictive process analytics aims to forecast future states of ongoing processes, such as remaining time or outcome prediction, leveraging event logs enriched with object-centric data. Current research emphasizes improving model interpretability and fairness through techniques like adversarial learning, counterfactual explanations (generated using methods such as evolutionary algorithms and Markov models), and Shapley values, alongside the application of graph neural networks and sequence-to-sequence models. This field is significant for enhancing the trustworthiness and usability of predictive models in business process management, enabling more informed decision-making and improved operational efficiency.