PMCube: A Data-Warehouse-Based Approach for Multidimensional Process Mining
Process mining provides a set of techniques to discover process models from recorded event data or to analyze and improve given process models. Typically, these techniques give a single point of view on the process. However, some domains need to different
- PDF / 1,174,865 Bytes
- 12 Pages / 439.37 x 666.142 pts Page_size
- 60 Downloads / 255 Views
Abstract. Process mining provides a set of techniques to discover process models from recorded event data or to analyze and improve given process models. Typically, these techniques give a single point of view on the process. However, some domains need to differentiate the process according to the characteristic features of their cases. The healthcare domain, for example, needs to distinguish between different groups of patients, defined by the patients’ properties like age or gender, to get more precise insights into the treatment process. The emerging concept of multidimensional process mining aims to overcome this gap by the notion of data cubes that can be used to spread data over multiple cells. This paper introduces PMCube, a novel approach for multidimensional process mining based on the multidimensional modeling of event logs that can be queried by OLAP operators to mine sophisticated process models. An optional step of consolidation allows to reduce the complexity of results to ease its interpretation. We implemented this approach in a prototype and applied it in a case study to analyze the perioperative processes in a large German hospital. Keywords: Data warehousing · OLAP · Multidimensional process mining · Comparative process mining
1
Introduction
Process mining comprises a set of techniques for the automatic analysis of business processes. It is based on events which are recorded during process execution and collected in an event log. Figure 1 illustrates the typical structure of an event log. Inside the event log, the events are grouped by their process instance or case. The ordered sequence of events belonging to a case is called trace. Both, events and cases, may have arbitrary attributes holding additional information about the observed events or process instances. Generally, the field of process mining can be classified into three kinds: (1) process discovery generates a process model describing the behavior recorded in the event log, (2) conformance checking compares an event log to a model in order to measure the quality of the model, and (3) process enhancement maps additional information stored in the events attribute (e.g., timestamps) to the process model to enrich it with new perspectives (e.g., execution times). c Springer International Publishing Switzerland 2016 M. Reichert and H.A. Reijers (Eds.): BPM Workshops 2015, LNBIP 256, pp. 167–178, 2016. DOI: 10.1007/978-3-319-42887-1 14
168
T. Vogelgesang and H.-J. Appelrath
Fig. 1. General structure of an event log
Process mining can be applied to healthcare processes. In contrast to traditional business processes, healthcare processes are typically unstructured and very complex, due to the individuality of patients. The treatment has to be adjusted to the individual situation of the patient considering age, sex, comorbidity, and so on. Institutional characteristics, like available resources or the experience of the medical staff, may influence the treatment process as well. For the analysis of such processes, it is desirable to minimize t
Data Loading...
