Risk models are a key competition factor in the business of Motor Insurance. Actuaries are looking for new techniques to make the most out of their data. The inclusion of modern Data Science methods into the actuarial daily work has proved to enhance the quality of risk models and to provide a better insight into the available data.
We will give a high level outlook on future development of (technical) Pricing in Motor Insurance and its impact on the work of actuaries. Data Science is an important part of this – our vision for the future of pricing.
After providing a typical Machine Learning workflow, we introduce the architecture of popular Data Science methods (Gradient Boosting, Neural Networks) and illustrate the strengths and weaknesses of these methods – also compared to GLMs. Further, we present approaches to evaluate and compare models to get a better understanding in terms of model interpretation but also in terms of identifying weak spots of currently used models.
Out of the demonstration of superiority of risk models enriched with Data Science methods, we finally provide some concrete business cases how to generate an added value for the insurance company.
Organised by the EAA - European Actuarial Academy GmbH.