Process simulation and pattern discovery through alpha and heuristic algorithms

Abstract

The paper is divided into two main parts. In the first part of the study, we applied two process mining discovery techniques (i.e., alpha and heuristic algorithms) on an event log previously collected from an information system during an Academic Writing (English) training course at a private university in Thailand. The event log was initially consisted of 330 process instances (i.e., number of participants) and 3,326 events (i.e., number of actions/tasks) in total. Using alpha algorithm enabled us to reconstruct causality in form of a Petri-net graph/model. By using heuristic algorithm we could derive XOR and AND connectors in form of a C-net. The results showed 86.36% of the applicants/participants managed to achieve the Academic Writing (English) certificate successfully, while 6.36% of them failed to achieve any certificate after a maximum number of 3 attempts to repeat the training course. Surprisingly, 7.28% of the participants neither achieved an accredited certificate nor failed the course by dropping out before ending the course training process. In the second part of the study, we used performance analysis with Petri net technique (as a process mining conformance checking approach) in order to further analyze the points of noncompliant behavior (i.e., so-called bottlenecks or points of noncompliant behavior) for every case in the collected course training log. Based on the results, we could eventually detect the existing discrepancies of the event log leading to +24 missed tokens and -24 remained tokens altogether.