Title: Multi-class classification approach to predict risk levels of insurance customers

Authors: Kaveh Khalili-Damghani; Mohsen Yaghoubizadeh

Addresses: Department of Industrial Engineering, South-Tehran Branch, Islamic Azad University, Tehran, Iran ' Department of Industrial Engineering, South-Tehran Branch, Islamic Azad University, Tehran, Iran

Abstract: In today's world, insurance business is one of the most important service industries. Attention to customers and understanding their behavioural patterns are carried out on the basis of different objectives such as reducing the pressure of competitors, increasing market share, reducing and optimising cost. Hence, the ability to understand and predict the behaviour and attitudes of customers has an important role in the success of insurance companies. In this study a multi-class classification approach is proposed to predict the behaviour pattern of insurance customers in a third-party car insurance company. The results of proposed approach are compared with those of traditional classification approaches. The results showed that proposed approach outperforms the traditional approach in the sense of providing high accuracy answers. The proposed approach of this study can be utilised in other classification problems in banks, libraries, hospitals, and chain stores.

Keywords: customer relationship management; CRM; individual classifiers; multiple classifiers; multi-class classification; ensemble learning; risk levels; insurance industry; behaviour patterns; customer behaviour; third-party car insurance.

DOI: 10.1504/IJMDM.2016.080707

International Journal of Management and Decision Making, 2016 Vol.15 No.3/4, pp.237 - 247

Available online: 28 Nov 2016 *

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