Initial steps to deploy and configure the software
Process Based Analytics
2 minute read
Failing business processes, like delayed delivery of cargo containers, can be costly to the entity conducting the business. To prevent business processes from failing and the costs associated with said failure, predictive business process monitoring predicts how an ongoing case will unfold. To this end, predictive business process monitoring uses the sequence of historical events occurred during the execution of a business case to make predictions about the future state of the case. If the predicted future state of the case indicates a problem, the ongoing case may be proactively adapted; e.g., by re-scheduling process activities or by changing the assignment of resources. The Process-based Analytics component aims at optimising the business process using machine learning techniques. A business process in the context of DataPorts can be the flow of vessels within the port’s service area or transport (containers or goods) operation process. By proactively predicting the future states of the ongoing process, the component provides forward-looking perspectives for the process managers to make decisions. It analyses business processes by using both historic and real-time data available inside the DataPorts platform to provide its predictive results to cognitive applications, which inform the end-users about the predictions. By exploiting advanced data analytics techniques and machine learning, these components offer decision support for terminal and process operators, thereby facilitating proactive management of port processes.
Ensemble Predictive Process Monitoring: process based analytics use ensembles of deep learning models (recurrent neural networks) to provide accurate predictions for each point during process execution, i.e., in a streaming fashion.
Prescriptive Process Monitoring: Building on process predictions, Online Reinforcement Learning allows automating the process on whether and when to adapt a running process. We apply state-of-the-art Reinforcement Learning algorithms to the problem of identifying the signs of possible failure early and accurately.
Explainable Predictive Process Monitoring: This feature provides interpretations on why a certain prediction is made by a black-box predictive model, in particular the deep learning models mentioned above. To generate highly accurate predictions and at the same time facilitate interpretability for predictive process monitoring tasks, we leverage the concept of model induction from interpretable machine learning (ML) research.
User guide to understand how the software is used
OpenAPI specification to interact programmatically
Links to the source Code available at eGitlab
Links to binaries and docker images for download