The categories have been defined in order to identify and differentiate between key methodology differences, for example, the choice between regression and precise interpolation and whether to build a formula from optimised components or optimise the complete formula. Further variations in model design are also being considered, including choice of basis function.
Key Objectives:-
To assess the impact of methodology and design choices against a number of relevant criteria, including accuracy, speed, complexity, ease of implementation and cost in order that a more formal understanding of these impacts can be documented and disseminated. Further investigations will attempt to improve our understanding of the models and their behaviours in order that improvements can potentially be made to the design, calibration and testing of proxy models as used in the insurance sector.
