Predictive Modeling Applications in Actuarial Science
 Volume 1
 Introduction
 Predictive Modeling Foundations
 Predictive Modeling Methods
 Bayesian and Mixed Modeling
 Longitudinal Modeling
 Volume 2
 Generalized Linear Model
 Extensions of the Generalized Linear Model
 Unsupervised Predictive Modeling Methods

Applications on Current Problems in Actuarial Science
 Chapter 8  The Predictive Distribution of Loss Reserve Estimates over a Finite Time Horizon
 Chapter 9  Finite Mixture Model and Workersâ€™ Compensation LargeLoss Regression Analysis
 Chapter 10  A Framework for Managing Claim Escalation Using Predictive Modeling
 Chapter 11  Predictive Modeling for UsageBased Auto Insurance
Chapter 6  Clustering in General Insurance Pricing
Authors
Ji Yao  University of Kent
JYao@uk.ey.com
Chapter Preview
Clustering is the unsupervised classification of patterns into groups (Jain et al., 1999). It is widely studied and applied in many areas including computer science, biology, social science and statistics. A significant number of clustering methods have been proposed in Berkhin (2006), Filippone et al. (2008), Francis (2014), Han et al. (2001), Jain et al. (1999), Luxburg (2007), and Xu and Wunsch (2005). In the context of actuarial science, Guo (2003), Pelessoni and Picech (1998), and Sanche and Lonergan (2006) studied some possible applications of clustering methods in insurance. As to the territory ratemaking, Christopherson and Werland (1996) considered the use of geographical information systems. A thorough analysis of the application of clustering methods in insurance ratemaking is not known to the author.
The purpose of this chapter is twofold. The first part of the chapter will introduce the typical idea of clustering and stateoftheart clustering methods with their application in insurance data. To facilitate the discussion, an insurance dataset is introduced before the discussion of clustering methods. Due to the large number of methods, it is not intended to give a detailed review of every clustering methods in the literature. Rather, the focus is on the key ideas of each methods, and more importantly their advantages and disadvantages when applied in insurance ratemaking.
In the second part, a clustering method called the exposureadjusted hybrid (EAH) clustering method is proposed. The purpose of this section is not to advocate one certain clustering method but to illustrate the general approach that could be taken in territory clustering. Because clustering is subjective, it is well recognized that most details should be modi ed to accommodate the feature of the dataset and the purpose of the clustering.