Learn, explore and unleash your inner chef.

What are the main consequences of machine learning on decision making in pricing?

Adopting machine learning solutions in the pricing sector has a big potential for taking the insurance industry to its natural next level. On one side finding the most sophisticated model of risk/loss allows for a more comprehensive understanding of our customers and therefore enhances fairness. By delivering on the promise of a premium tailored to the convergent needs of the company and the client, we aim at finding the most appropriate solution for both parties. With the growing amount of data available for segmentation, it is just through advanced statistical methods that we can find our way in this information deluge. On the other side machine learning techniques allow for dynamic pricing, therefore updating the available information on the fly, and allowing for the identification of the low hanging fruit in a constantly evolving market.

Will machine learning kill GLM or complement it?

Generalised Linear Models, as the name suggests, are a specific class of models that generalised the less flexible linear regression models. They did not replace simpler models: they rather built on the pre-existing know-how and developed further in complexity. In a similar way, various machine learning techniques allow for increased level of flexibility, both in how we understand what data tell us, and in how we think of which questions we want answered. The general approach though, is to let the data speak first, remembering the motto by Box “all models are wrong, but some are useful”. On top of that, the big players in the insurance market have naturally a bigger inertia, when it comes to implementing innovative solutions across all teams, so GLMs are going to be around for years side by side to new solutions.

Could you please explain the new competencies required to understand machine learning?

This question relates back to previous one, in that the modern actuary needs data scientist skills to complement the more traditional competences of our profession. Alternatively, some teams prefer to structure themselves in a way that synergically puts actuaries and data scientists in the same environment. Someone who is skilled in adopting machine learning solutions in the pricing sector has been recently described as a distillation of statistics, computer science, and business analytics. Which means that this profile is typically a better programmer than a statistician, and has a deeper analytical understanding of the technical solutions than a business analyst.  This is just an example, but it conveys the idea of the need of more hybrid figures in teams that want to keep an edge on the market.

What would you like to achieve by attending the Pricing in Personal Lines Insurance?

We want to deliver attractive products with attractive prices to our customer and of course we are also in the business to generate profits in a sustainable way. To achieve both aims, we use advanced pricing techniques.