|
Abstract:
|
Regression models are a useful type of variable predictions that can be used for automobile insurance pricing calculations. Due to the frequent presence of non-linear relations, the use of semi parametric regression can improve the pricing prediction. In this presentation we show the results in a table as function of automobile and policy-holders variables. Our proposed methodology consists of a nonparametric smoothing curve fitted using penalized splines, together with a mixed model which have been useful in solving the problem. The predictions are important because they complete the pricing process. That is, when a new client reports inputs which have not been tabulated by the insurance company before, a price for the insurance cost can be found through the model.
|
