What are the Basic Similarities and Differences between Insurance Risk Modelling and Standard Machine Learning Techniques?

In industry, it is regularly encountered that there is some confusion about the similarities and differences between insurance risk modelling and standard machine learning techniques. Hence, we have developed simple overview table which helps to eliminate most of these ambiguities. In this talk, we discuss this table while covering quantitative and qualitative aspects. The talk will consist only of a single slide for the whole talk. The goal of the presentation is to help actuaries, data scientists, management and insurance specialists to better understand the conceptual commonalities and differences between the two disciplines. The table shall also serve to decide which modelling approach is appropriate for a given modelling challenge.

This work has been done as part of the “Data Science” working group of the Swiss Association of Actuaries (SAA). The group publishes material that discuss the use of machine learning techniques for actuarial applications, see the group’s website www.actuarialdatascience.org for further information.