Title: Regression and decision tree approaches in predicting the effort in resolving incidents

Authors: Sharon Christa; V. Suma; Uma Mohan

Addresses: Department of Information Science and Engineering, Dayananda Sagar College of Engineering, Bangalore, India ' Research and Industry Incubation Center, Department of Information Science and Engineering, Dayananda Sagar College of Engineering, Bangalore, India ' London School of Commerce, Chaucer House, White Hart Yard, London SE1 1NX, UK

Abstract: IT service management plays a key role in software maintenance. Service management offers the customers a platform to raise the incidents that needs to be resolved. This papers is a comprehensive analysis performed on research in the area of production support. Lacunas are identified in different areas of production support services. The necessity of a generalised proactive model that can predict the effort required in closing incident tickets are identified. The paper further presents the scope of integrating machine learning approaches to predict effort in an incident management system of production support. Two different approaches are considered in modelling namely, regression-based and tree-based modelling. In tree-based modelling, basic decision tree and random forest models are used along with multiple linear regression model. In order to build the model, real-time dataset is used. The models are verified using a real-time test dataset. The models being dataset dependent did not generalise and converge well due to which, the possibility of developing other models using different machine learning techniques are discussed.

Keywords: production support; effort prediction; decision tree; random forest; multiple linear regression; incident management; tickets; incident logs; ITIL; time to resolve.

DOI: 10.1504/IJBIS.2022.122342

International Journal of Business Information Systems, 2022 Vol.39 No.3, pp.379 - 399

Received: 03 Nov 2018
Accepted: 05 Aug 2019
Published online: 21 Apr 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article