Defining a general approach for business process analysis that helps to better understand and support quality of business process outputs

Many organizations tend to use Workflow-Management Systems (WfMS) to increase efficiency and quality of business processes. If processes deviate from acceptable behaviour, they become exceptional. Process owners often know about occurring exceptions, but they often do not have the time and the know-how to define improvement measures that would reduce the occurrence of workflow exceptions. The goal of this study was to define a general approach for business process analysis that helps to support quality of business process outputs. The output of the approach were improvement measures that, when applied to a specific business process, should have a positive impact on process quality by reducing occurring workflow exceptions. This thesis is based on data obtained from the Swiss National Bank’s (SNB) introduction process for new employees. The analysis of data combined with interviews showed that several exceptions with different characteristics occur during process runtime. To identify different exception origins, a data model out of WfMS data enriched with contextual information and constructed attributes has been defined. The data model has been developed using the CRISP-DM reference model of (Chapman et al., 2000). Afterwards learning algorithms have been applied to define a data model to identify data patterns that lead to exceptions. Resulting data patterns have been analysed to derive improvement measures. Based on made experiences a general approach to improve process quality has been defined.