Decision Mining in the Rail Industry - A Case Study on an Application of Decision Mining for Process Model Enhancement and Conformance Checking in the Context of an Industrial Wheelset Revision Process

Abstract

The field of process mining has evolved into a multidisciplinary research paradigm in which relevant data mining practices led to new avenues of analysis and better process understanding. However, the role of decisions in relation to the modeling and analysis of processes is still significantly unexplored. The nascent discipline of decision mining aims to fill this void by separating concerns, while a more holistic approach is also being considered. This research took the latter approach, as there is nowadays an abundance of process data available that can potentially uncover captivating details without the need of a separate project. Following the design science research methodology, a methodological framework for integrated process and decision mining was developed. It is based on the synthesis of an established process mining project methodology and a review of the state-of-the-art in decision mining approaches. The extended methodology was applied and evaluated with a case study of the industrial wheelset revision process at the Dutch national railway company, which included a focus group with several experts. The results demonstrated that the addition of a decision perspective to process models allows for better process understanding and therefore is potentially useful for an organization that is executing a process mining project. In addition, the evaluation identified a new form of conformance checking that can be used to validate whether the process was executed correctly in accordance with the decisions taken. Future work could aim to increase generalizability by applying the framework in other contexts. Furthermore, the framework could be validated further beyond the sole evaluation of the resulting artifacts and insights.