Predictive Process Monitoring using a Markov Model Technique

Information systems play a vital role in business process understanding through recording data where each process execution is represented in event logs. These event logs consist of a collection of traces which characterise the execution of a process. The techniques that analyse such types of data are broadly known as process mining. Hence, Process mining is a family of techniques to evaluate and analyse business process based on their pragmatic behaviour as recorded in event logs. Predictive process monitoring aims to predict how the completion of running process events can be anticipated. In this paper, Markov chain models have been investigated for prediction of future process events by considering a sequence of events. The Markov model is a special type of statistical (process) model that are used to evaluate systems where it is considered that future states depend only on the current state and not on previous states. Results have been evaluated using an accuracy metric for a dataset contains tasks from a ticketing management process of a software company.