B1656
Title: Regression trees for extreme events, applications to natural disasters and cyber insurance pricing
Authors: Olivier Lopez - Ensae IP Paris (France) [presenting]
Abstract: A regression tree procedure adapted to the analysis of extreme events is introduced. We show theoretical results assessing the performance of the procedure for finite sample size. We then show how this tool can be used to build priors for bayesian credibility pricing in insurance. Two applications are considered, one in a natural disaster, and the other in cyber insurance. In each case, the regression tree approach allows linking a claim to a risk class, in order to improve the information conveyed by historical data from the victim.
