SITUATION

Our client has at disposal a lot of data from the bank environment. We wish to use these data in order to create a new insurance tariff for the motor third-party liability insurance with only bank information.

SOLUTIONS

Our solutions: Preselection of variables for the frequency and the costs via LASSO, Random Forests and GBM. Creation of a version of the Genetic Algorithm. Split between attritional and large claims with binomial link between segments.