Process mining analysis starts with an event log. Such a log can be obtained from various sources, such as a workflow system, a ERP system, or even from a web log. Importantly, the richness of the information recorded in a log has an impact on the accuracy and reliability of process mining analysis.
Insights about the performance of one’s processes are typical valuable outcomes of a process mining analysis. Nevertheless, despite an increased uptake of process mining appllications in recent years as evidenced by success stories and increased tool support, current performance analysis capabilities remain somewhat limited in the context of information-poor event logs.
The goal of this research is to develop a new framework, and its corresponding tool, to facilitate sophisticated performance-related analysis of event logs containing varying degrees of information (from information-poor to information-rich logs). In particular, we look into the concept of event gap as the basic building blocks for the new performance analysis framework.
The main research question we attempt to answer in this research is “What types of performance-related information can we extract from an information-poor log?”. Further, once we know the type of information that we can extract, we also need to investigate “to what extent can we exploit this information to provide deeper insights about the performance of a process?”.
This research was conducted in the context of Risk-aware Business Process Management project. Our research has shown that, even with an information-poor log, we can still obtain interesting performance-related information, including
- estimations of various waiting and working times (both from process perspective and resource perspective)
- estimations of resources workload at any given point in time
- insights into the performance cycles of processes and resources
- insights into how various factors, as recorded in an event log, may impact the duration of working/waiting times
We have realised the results of our research as a plug-in to the open-source ProM Tool, called “Event Gap Analysis”. For more information, please refer to the following journal article:
Suriadi, Suriadi, Ouyang, Chun, van der Aalst, Wil M.P., and ter Hofstede, Arthur H.M. (2015), Event interval analysis: Why do processes take time? Decision Support Systems, 79, pp. 77-98.