An Approach to E ort Reduction in the Incident Management System During Software Maintenance Using Machine Learning Techniques

Abstract

One of the most desired capability of any IT service industry is to ac curately estimate the e ort required to resolve an incident. Accurate estimation of e ort not only help account managers to take relevant de cisions but also, to plan the resource utilization. The e ort estimation techniques that exists in the industry is based on domain knowledge, ex perience of the account managers and regression based techniques which lacks accuracy. Further, each ticket accelerates the overhead to resolve the incident since e ective ticket resolution process plays an in uencing role in customer satisfaction. Closure of tickets in compliance with SLA newlineis always a challenge when it comes to unknown parameters. Hence, ef fective incident management involves better insight on the e ort required in resolving an incident and closing the associated ticket. Therefore the main objective of this research is newline To analyze the signi cance of incidents when compared to defects in the e ective service management. newline To identify the factors that has direct impact in the estimation of resolution e ort. newline To introduce high precision predictive models for e ort reduction in the production support domain. newline newline