Effective Statistical Learning Methods for Actuaries I

This book summarizes the state of the art in generalized linear models (GLMs) and their various extensions: GAMs, mixed models and credibility, and some nonlinear variants (GNMs). In order to deal with tail events, analytical tools from Extreme Value Theory are presented. Going beyond mean modeling, it considers volatility modeling (double GLMs) and the general modeling of location, scale and shape parameters (GAMLSS). Actuaries need these advanced analytical tools to turn the massive data sets now at their disposal into opportunities.

The exposition alternates between methodological aspects and case studies, providing numerical illustrations using the R statistical software. The technical prerequisites are kept at a reasonable level in order to reach a broad readership.

This is the first of three volumes entitled Effective Statistical Learning Methods for Actuaries . Written by actuaries for actuaries, this series offers a comprehensive overview of insurance data analytics with applications to P&C, life and health insurance. Although closely related to the other two volumes, this volume can be read independently.

Features numerous examples and case studies in P&C, Life and Health insurance Provides a broad and self-contained presentation of insurance data analytics techniques, from classical GLMs to neural networks Addresses many specific issues which arise in insurance data analysis Can be used as course material, for CPD programs or for self-study Complements the existing literature on GLMs in insurance Written by actuaries for actuaries Based on more than a decade of lectures and consulting projects on the topic, by the three authors

Auteur
Michel Denuit holds masters degrees in mathematics and actuarial science as well as a PhD in statistics from ULB (Brussels). Since 1999, he has been professor of actuarial mathematics at UCLouvain (Louvain-la-Neuve, Belgium), where he serves as Director of the masters program in Actuarial Science. He has also held several visiting appointments, including at Lausanne (Switzerland) and Lyon (France). He has published extensively and has conducted many R&D projects with major (re)insurance companies over the past 20 years.

Donatien Hainaut is a civil engineer in applied mathematics and an actuary. He also holds a masters in financial risk management and a PhD in actuarial science from UCLouvain (Louvain-La-Neuve, Belgium). After a few years in the financial industry, he joined Rennes School of Business (France) and was visiting lecturer at ENSAE (Paris, France). Since 2016, he has been professor at UCLouvain, in the Institute of Statistics, Biostatistics and Actuarial Science. He serves as Director of the UCLouvain Masters in Data Science.

Julien Trufin holds masters degrees in physics and actuarial science as well as a PhD in actuarial science from UCLouvain (Louvain-la-Neuve, Belgium). After a few years in the insurance industry, he joined the actuarial school at Laval University (Quebec, Canada). Since 2014, he has been professor in actuarial science at the department of mathematics, ULB (Brussels, Belgium). He also holds visiting appointments in Lausanne (Switzerland) and in Louvain-la-Neuve (Belgium). He is associate editor for the Journals Astin Bulletin and Methodology and Computing in Applied Probability and qualified actuary of the Institute of Actuaries in Belgium (IA|BE).

Contenu
Preface.- Part I: LOSS MODELS.-1. Insurance Risk Classification.-Exponential Dispersion (ED) Distributions.-3.-Maximum Likelihood Estimation.-Part II LINEAR MODELS.-4. Generalized Linear Models (GLMs).- 5.-Over-dispersion, credibility adjustments, mixed models, and regularization.-Part III ADDITIVE MODELS.- 6 Generalized Additive Models (GAMs).- 7. Beyond Mean Modeling: Double GLMs and GAMs for Location, Scale and Shape (GAMLSS).- Part IV SPECIAL TOPICS.- 8. Some Generalized Non-Linear Models (GNMs).- 9 Extreme Value Models.- References.