Description

The artificial intelligence (AI) industry has been growing rapidly over the past few years, to such an extent that some now class it as an industry, no longer “emerging” but clearly “emerged.” As we enter 2020, it seems like the perfect opportunity to assess the industry as a whole and in particular how it affects the design and management of insurance products.

At the heart of these developments are a number of basic data science and machine learning modelling ‘techniques and tools’ that actuaries need to understand in order to keep up to date. Big data, predictive models and automated decision making have disrupted the insurance new business, underwriting and risk management processes over the past few years. How did we get here? What does the future hold?

This year’s International Summer School of the Swiss Association of Actuaries (ISS2020) will review the lessons learned from the evolutionary process and discuss what the potential next steps are in this journey. We will address for instance the following questions: “Are real-time fully underwritten decisions an achievable goal?” and “How can risk management decisions be learned automatically through machine learning?”

We will start from ground zero, reviewing basic techniques already included in classical data science textbooks such as “An Introduction to Statistical Learning; with Applications in R”, by G. James, D. Witten, T. Hastie and R. Tibshirani, published by Springer in 2013. As no single textbook addresses fully the insurance applications of these techniques, over the course of a week we will cover a selection of the most important machine learning methods for actuaries. The second portion of the course will be dedicated to life insurance risk models and their inclusion in hedging procedures for life insurance products tied to investment returns. Traditional hedging procedures shall be reviewed, followed by an introduction to reinforcement learning which allows learning optimal hedging policies from experience. No prerequisites are required other than an intermediate level knowledge of Statistics and the use of databases as well as statistical software.

The emphasis will be on insurance applications, both in life and non-life, and learning through examples. The teachers will be joined by Alexandre Carbonneau who will help with the insurance illustrations, exercises and illustrative scripts in the open-source R language.