Boosting Poisson regression models with telematics car driving data

Abstract With the emergence of telematics car driving data, insurance companies have started to boost classical actuarial regression models for claim frequency prediction with information. In this paper, we propose two data-driven neural network approaches that process data complement pricing a behavior risk factor from data. Our networks simultaneously accommodate feature engineering and modeling which allows us integrate in one-step approach into models. We conclude our numerical analysis both factors are necessary receive best predictive This emphasizes these sources information interact each other.