Wildfire Loss Modeling: A Flexible Semiparametric Approach (Q6953013)

The paper is devoted to the fitting of a model for fire-loss in the USA. Authors refer to a first approach by using a fitting of a mixed distribution of single claims. After this way, they fit a time-series model whose every single fitted variable comes from an actually a mixed-logistic distribution. This model is actually semi-parametric, since the distribution of the `explanation variable' is not pre-specified. The authors also provide a spline approach related to the interpolation of between the values of the variables which may be included in a similar model. This approach is a way of ``flexible-regression'' in cases where the values of the `explanation variables' in some intervals are well-known, while the fitting of a regression model needs intermediate values of them. The authors' treatment refers to time-series models, but this approach may be used in any regression model.