We provide a tutorial that compares a classical generalized linear model for claims frequency modeling to regression tree, boosting machine and neural network approaches. We explore these methods, discuss their calibration and study their predictive power on an explicit motor third-party...