IT Incident Solving Domain Experiment on Business Process Failure Prediction

Abstract

Business process monitoring aims at maintaining the reliability of process executions. Nevertheless, the dynamic nature of business processes hinders a proactive scenario in which risk mitigation actions can occur before the facts that put the process at risk. We argue that understanding failures behavior allows proactive actions. Analysing historical data of processes executions supports the identification of situations and patterns of failure behavior. In this paper, we present an experiment in which a combination of well-established techniques from Data Mining and Process Mining fields are applied to an incident management process. The results obtained show that it is possible to identify failures in order to reach for a proactive risk mitigation scenario.