Quantifying Uncertainty in Actuarial Models: An Introduction to Conformal Prediction
As the actuarial landscape becomes increasingly data-driven, traditional statistical methods are evolving, linking point estimates with quantifiable uncertainty. Conformal Prediction is a powerful framework used to evaluate the uncertainty of predictions. It turns point predictions into prediction regions, in this way, when you make a prediction, the output has probabilistic guarantees that it covers the true outcome. In this web session, we will explore the theoretical foundations of Conformal Prediction, its assumptions, methodology, and advantages over traditional approaches. Participants will see how this technique can be a disruptive innovation in risk assessment, pricing, reserving, and forecasting by integrating uncertainty directly into predictions. The session is balanced between theory and hands-on activities, offering real-world examples and code demonstrations in classification, regression, time series, natural language processing (NLP), and computer vision, all applied to challenges in the actuarial field.

• Actuarial analysts and consultants seeking to enhance their understanding of model uncertainty
• Data scientists working in insurance or finance who want to improve the reliability of their predictive models
• Actuarial managers and directors interested in more robust and transparent risk management
• Anyone with a basic knowledge of statistical modelling and Python.

Participants should have access to Google Colaboratory.

Technical Requirements
Please check with your IT department if your firewall and computer settings support web session participation (the programme Zoom will be used for this online training). Please also make sure to join the web session with a stable internet connection.

The web session aims to equip actuarial professionals with the knowledge and skills behind Conformal Prediction, improving the reliability and interpretability of their predictive models. By the end of the course, participants will understand how to apply conformal methods to various types of data and leverage these techniques to improve decision-making processes in actuarial tasks.

The language of the web session will be English.

Claudio Giorgio Giancaterino
Claudio Giancaterino is a qualified Actuary who works during the day with Intesa Sanpaolo Assicurazioni, an Italian Insurance Company based in Milan. In his free time, he is involved in data science activities. In the past, he was an assistant professor of Insurance Statistics at the Catholic University of Milan. He collaborated with the IAA and IFoA in several working parties and attended several data science competitions. He held workshops and talks at several conferences/meetups. He’s a member of the Astin and the Italian Actuarial Body Association.