DATA MINING APPROACH TO MODELLING POLICYHOLDER'S CLAIM BEHAVIOUR
Anna Kohlmayr, Nick Scerbakov · Knowledge management · 2007
AbstractThe goal of this paper is to demonstrate how the techniques of data mining can be applied in personal accident insurance. Specifically, a personal accident claims database of a major national insurance company is analysed to aid in pinpointing patterns in policyholders' claim behaviour and evaluating claim trends. The policyholder's claim behaviour is described in terms of the major parameters of the claim sequence model. To support insurer's decision-making, we also consider the probability aspect of the claim distribution in the personal accident insurance. Furthermore, the ratio of premium revenues to claim payments is examined with the purpose of exploring the potential for improvement in the claim handling process. This paper suggests a novel approach to modelling policyholder's claim behaviour. Using this approach, scalable and effective measures for analysis and monitoring insurance customer's claim behaviour are introduced. Non trivial results are obtained. The proposed model and the results are empirically verified. Application of the results of this type of analysis is presented.