Business Process Control-Flow Complexity Prediction Using Machine Learning-based Tool

Business processes are structured sequences of activities, events, and other elements designed to fulfill customer needs in services or goods. Business process management is an approach used to combine information technology and management fields to develop, maintain, and improve organizational workflows. Business process modeling is a key technique employed for graphical description and analysis of organizational activities, it helps to understand, capture, analyze, and improve enterprise business processes. Thus, the quality of designed models should be carefully controlled to ensure timely business process modeling errors detection and prevention prior to they turn into errors leading to extra costs or other negative impacts on an organization. Business Process Model and Notation (BPMN) is the de-facto standard of workflow description in recent years, providing a well-structured notation to describe complex organizational activities. Hence, in this study we consider the complexity of BPMN models as one of the main quality attributes formulated in the business process modeling field. The task of complexity evaluation of BPMN models is addressed, the key complexity metrics of BPMN models are considered, a prediction machine learning model is proposed using linear regression model, the respective software tool is designed, implemented and verified. Obtained experimental results are analyzed and discussed.