Predicting Remaining Cycle Time from Ongoing Cases: A Survival Analysis-Based Approach

Predicting the remaining cycle time of running cases is one important use case of predictive process monitoring. Different approaches that learn from event logs, e.g., relying on an existing representation of the process or leveraging machine learning approaches, have been proposed in literature to tackle this problem. Machine learning-based techniques have shown superiority over other techniques with respect to the accuracy of the prediction as well as freedom from knowledge about the underlying process models generating the logs. However, all proposed approaches learn from complete traces. This might cause delays in starting new training cycles as usually process instances might last over long time periods of hours, days, weeks or even months. In this paper, we propose a machine learning approach that can learn from incomplete ongoing traces. Using a time-aware survival analysis technique, we can train a neural network to predict the remaining cycle time of a running case. Our approach accepts as input both complete and incomplete traces. We have evaluated our approach on different real-life datasets and compared it with a state of the art baseline. Results show that our approach, in many cases, is able to outperform the baseline approach both in accuracy and training time. © 2021, Springer Nature Switzerland AG.