From Astroparticle Physics to Insurance Tech – Data Science at Zurich

What does the price of your car insurance have to do with modeling a blazar spectrum? At the first glance, not much. But beneath the surface, both problems involve extracting information from complex and (often times messy) data. Insurance premiums are priced on the risk and cost of potential incidents. When you consider how these costs are calculated, similarities to physics modeling become clear: First, you need reliable data sources, processing pipelines, and quality monitoring. Next, you deal with incomplete or sparse data, determine the best observables, and fit robust models. Finally, you use your models to forecast what to expect in the future. And while the data itself might be different, the challenges faced in physics and insurance modelling are surprisingly similar. In this talk I am sharing how data science and data engineering make a difference in the insurance industry.