IT Help Desk Incident Classification using Classifier Ensembles

Proper assignment of IT incident tickets raised by the end users is a very crucial step in an IT Service management system. Incorrect manual selection of incident category while raising the ticket causes assignment of incident to a wrong domain expert team which in turn results in unnecessary resolution delay and resource utilization. In this work, we proposed machine learning based model for auto categorization of incident category by mining the user’s natural language description of the incident. Classification techniques such as Naive Bayes and Support Vector Machines are used as base classifiers to model the incident classifier system. To further analyse the classifier performance we used the ensemble classifier techniques such as Bagging and Boosting to build the incident classifier model. The performance of base classifiers and ensemble of classifiers are analysed using various performance metrics. Ensemble of classifiers outperformed well in comparison with the corresponding base classifiers. Pre-processing of the IT incidents description data is one of the key challenges in this research work due to its unstructured nature. The proposed automated incident classification model results in simplified user interface, faster resolution time, improved productivity and user satisfaction and uninterrupted flow in business operations. The real world IT infrastructure incidents data from a reputed enterprise is used for our research purpose.