Predictive Modeling Applications in Actuarial Science

Chapter 3 - Generalized Linear Models as Predictive Claim Models


Greg Taylor | University of New South Wales

James Sullivan

Chapter Preview

The present chapter is concerned with a dataset that is in fact far from conformity with the chain ladder assumptions. Its modeling requires a great deal more structure than provided by the chain ladder, so that informed decision-making will be achieved only by means of statistical methods. The chapter is concerned with the physical circumstances accounting for the breach of chain ladder assumptions, the recognition of these circumstances in the model formulation, and the representation of the resulting model in the form of a generalized linear model (GLM).