Why High Dimensional Modeling in Actuarial Science?

We describe the Generalized Additive Family which covers Generalized Linear Model, Generalized Additive Model, Neural Network and Boosting and more. We outline the theoretical features of each member, along with the corresponding implications on actuarial applications. A complete procedure that details the data-scrubbing, preliminary data diagnosis, variable selection, model selection and model diagnosis is presented for each family member to model claim frequency for a set of real life data. This paper aims to provide a robust framework to actuaries on how multiple modeling techniques can help enhance the credibility of popular pricing models.