Insights from Inside Neural Networks

64 Pages Posted: 19 Aug 2018 Last revised: 24 Apr 2020

See all articles by Andrea Ferrario

Andrea Ferrario

Dep. Management, Technology, and Economics ETH Zurich; Mobiliar Lab for Analytics at ETH

Alexander Noll

PartnerRe Ltd - PartnerRe Holdings Europe Limited

Mario V. Wuthrich

RiskLab, ETH Zurich

Date Written: April 23, 2020

Abstract

We provide a tutorial that illuminates the aspects which need to be considered when fitting neural network regression models to claims frequency data in insurance. We discuss feature pre-processing, choice of loss function, choice of neural network architecture, class imbalance problem, balance property and bias regularization as well as over-fitting. This discussion is based on a publicly available real car insurance data set.

Keywords: Neural Networks, Architecture, Over-Fitting, Loss Function, Dropout, Regularization, LASSO, Ridge, Gradient Descent, Class Imbalance, Car Insurance, Claims Frequency, Poisson Regression Model, Machine Learning, Deep Learning

JEL Classification: G22, C10, C13, C14, C67

Suggested Citation

Ferrario, Andrea and Noll, Alexander and Wuthrich, Mario V., Insights from Inside Neural Networks (April 23, 2020). Available at SSRN: https://ssrn.com/abstract=3226852 or http://dx.doi.org/10.2139/ssrn.3226852

Andrea Ferrario

Dep. Management, Technology, and Economics ETH Zurich ( email )

Zurich
Switzerland

Mobiliar Lab for Analytics at ETH ( email )

Zürich, 8092
Switzerland

Alexander Noll

PartnerRe Ltd - PartnerRe Holdings Europe Limited ( email )

160 Shelbourne Road
Dublin, 4
Ireland

Mario V. Wuthrich (Contact Author)

RiskLab, ETH Zurich ( email )

Department of Mathematics
Ramistrasse 101
Zurich, 8092
Switzerland

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
2,486
Abstract Views
7,025
Rank
10,643
PlumX Metrics