Insurance Data Science Conference


The third annual Insurance Data Science Conference was held from June 16 through 18, 2021 via Zoom. There were over fifty talks with two keynote speakers, a panel discussion and a platform called Gather “that allowed delegates to walk around in virtual rooms to meet others. As soon as your avatar would get close to someone else, a little video chat window would open to get the conversation started,” explained Markus Gesmann a mathematician working in insurance and capital markets who is the founder of the Insurance Data Conference series. If you weren’t one of the 250 attendees, you can find all the presentation files here.

Keynote Speakers

One of the keynote speakers was Thomas Wiecki, CEO of a Bayesian consultancy and author of PyMC3, a probabilistic programming framework written in Python. While his problem was presented through a rocket launch product he was highlighting how PyMC3 can be used to balance various trade-offs such as risk uncertainty, correlations and upside to find the optimal allocation of an insurance budget.

The other keynote speaker was Bettina Grun an applied statistician at WU Vienna, Austria – the University of Economics and Business. She presented an overview on recent advances in model-based clustering including challenges due to big data settings, new model extensions as well as developments regarding estimation and validation methods.

Panel Discussion

Led by moderator Trevor Maynard, Head of Innovation at Lloyd’s, the panel discussion explored how to harness the best of new AI technology and control the risks of it. As you are well aware there are still many open ethical questions about AI and the European Union published their first proposal to regulate AI shortly before the conference.

Individual Presentations

The 54 talks were very technical, as would be expected, but also very well distributed across topics of interest to actuaries and insurance data science specialists. Claims presentations had the highest representation and included Enhancing Auto Claim Review using Machine Learning and Gamma Mixture Density Networks and their application to modeling insurance claim amounts. Neural Networks were presented in six different talks, while Deep Learning, AI and Bayesian methods all were discussed multiple times. You’ll be happy to know COVID-19 only came up in two talks and climate change made it to one.


Swiss Re Institute and Mirai Solutions sponsored this year’s conference which was delayed a year when the 2020 conference was cancelled. It is the third in the insurance data science conference series after Zurich in 2019 and London in 2018. Mark your calendars for the fourth insurance data science conference in Milan, Italy on June 15-17, 2022.

Interested in hiring or discussing your career as an insurance data scientist? Contact Smith Hanley Associates’ Actuarial Recruiter, Rory Hauser at

Leave a Reply

Your email address will not be published. Required fields are marked *