Parametric and XGBoost Hurdle Model for estimating accident frequency

Dafnis Krasniqi (1) , Jean-Marc Bardet (1) , Joseph Rynkiewicz (1)

1 SAMM - Statistique, Analyse et Modélisation Multidisciplinaire (SAmos-Marin Mersenne)

The Poisson model is a commonly used method for modeling count data with exogenous variables, but it can be limiting when dealing with data that has a high proportion of zeros. To address this issue, we propose the use of the Hurdle model as an alternative approach. We discuss the properties of the Hurdle model and demonstrate how it can be implemented using both parametric and nonparametric estimates, including the XGBoost method. To evaluate the effectiveness of our proposed XGBoost Hurdle model, we apply it to a car insurance dataset from a French insurance company. This dataset includes a significant number of drivers with zero accidents per year. Our results show that the XGBoost Hurdle model outperforms several other models when applied to this data type.

Mots clés: Insurance pricing; XGBoost algorithm; GLM; Hurdle Model

Domaines: Statistiques [math.ST]

Dafnis Krasniqi, Jean-Marc Bardet, Joseph Rynkiewicz. Parametric and XGBoost Hurdle Model for estimating accident frequency. 2023. ⟨hal-03739838v2⟩