Predicting Performances in Business Processes using Deep Neural Networks

Abstract

The starting point of the business process improvement is to identify where the process can be improved. Process mining has played an instrumental role in this regard. However, existing approaches in the process mining discipline assume that a business process is in a steady state and make a “snapshot” of the process performance. Instead, a business process evolves due to a sequence of random events or a gradual change in the process. This dynamic nature of the business process requires a new approach to identify patterns in the evolution and make predictions of future performances based on the patterns. The predictions enable business managers to take proactive actions in real-time to improve the process and mitigate risks. In this paper, we propose a novel method to predict the future performances of a business process from the historical records of the performances. More in detail, we construct an annotated transition system and generate a process representation matrix from it. Based on the process representation matrix, we build performance prediction models using deep neural networks which consider both spatial and temporal dependencies present in the underlying business process. To validate the proposed method, we performed two case studies, each of which is related to a healthcare service process and a manufacturing process respectively.