ML Explainability in Actuarial Data Science: A Practical Primer

Announcement from the EAA organiser:

These days, nobody disputes the profound impact and yet-untouched potential of Machine Learning and Artificial Intelligence anymore. Yet, in the actuarial sciences, these breakthrough possibilities are hampered by regulation, the need for numerical confidence and insight into model decision making, the latter being subsumed as a "black box problem". Thus, the quest for explainability is in fact much more pressing than in any other industry.

This upcoming seminar on ML explainability aims to provide insights into the areas of unsupervised learning, supervised learning and artificial neural nets via model-agnostic explainability approaches, while providing opportunities to try out the methods yourself!

Participants are expected to have, apart from actuarial basics, a preliminary grasp on Machine Learning and hypothesis testing as well as the R and/or Python languages.

Explainability is the cornerstone of practicability of ML and AI techniques – without thorough insights into the inner workings of the more advanced decision making and prediction methods, usage become hampered by the worst kind of information asymmetry: Unpredictable model outputs not fit for productive use. Ultimately, AI and ML methods will be well-suited for use in actuarial contexts, but to this end, more and better tools to understand these tools need to be procured and utilised.

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).

Click here to make a reservation: Your early-bird registration fee is € 600.00 plus 19% VAT for bookings by 16 January 2023. After this date, the fee will be € 780.00 plus 19% VAT.