Investigating the impact of collaborative patterns on business process performance: A large scale emprical study perspective

Identifying causal factors for process performance is critical to business process success. Therefore this research aims to investigate the impact of collaboration patterns on process performance in considering that process is a collaboration task. To make real life sense, we adopt a business process event log, Volvo log provided by BPIC 2013 as relevant data to conduct an empirical study for this impact. The log used here has a large scale of collaboration patterns and faces with unbalanced samples problem, thus in this paper, to overcome computation complexity resulted from large scale collaboration patterns, problem that the number of patterns is very large relative to samples and problem of unbalanced samples, we developed a methodology for investigating the impact of collaboration patterns on process performance. The methodology is a combination of logistic regression model which can handle unbalance samples problem easily, Stochastic Gradient Descent (SGD) which is efficient in large scale machine learning problems. It is expected that this research provided by us contribute to both business process management area and large scale empirical study in many domains.