The increasing use of artificial intelligence (AI) and machine learning (ML) in the insurance industry in general and in actuarial issues in particular presents both opportunities and risks. Acceptance of complex methods requires, among other things, a degree of transparency and explainability of the underlying models and the decisions based on them.
Welcome to this four-part training. In the first part, we will explore the concept of explainable artificial intelligence (XAI) through a qualitative discussion. We will not only characterize both model complexity and explainability, but also examine when a model can be considered sufficiently explained. Actuarial diligence will be addressed as well, using the counterfactual XAI method as an example. Additionally, we will provide an illustrative and comprehensive overview of explainability techniques, along with a compilation of useful and practical notebooks.
The second block will introduce the participants to variable importance methods. These methods try to provide an answer to the question: “Which inputs are the most important for my model?”. We will provide a general overview of variable importance methods and introduce some selected methods in depth. In addition to providing examples and use cases, we will cover enough of the theory underlying the methods to ensure that users have a good understanding of their applicability and limitations. Throughout, we will also discuss practical aspects of actuarial diligence such as how to interpret and communicate results from these methods.
In the third part, we will focus on further specific standard methods for XAI. Here, we explain how the model-agnostic methods “Individual Conditional Expectation”, “Partial Dependence Plot” and “Local Interpretable Model-Agnostic Explanations” work and refer to well-known Python packages and several Jupyter Notebooks. Additionally, we examine the model-specific tree-based “Feature Importance” of the Python package “scikit-learn”. Throughout this part, we also discuss aspects of actuarial diligence and limitations of the considered methods.
The last part of the online training provides an interactive, hands-on experience with explainable AI using a Jupyter notebook designed around an actuarial use case. Participants will be guided through a comprehensive machine learning workflow before exploring the implementation of various XAI techniques. In analyzing several XAI methods, we will study their main ideas on a conceptual level and their concrete implementations, apply each to the given machine learning problem, and discuss their advantages and disadvantages. The interactive segment concludes as participants are given an additional case study to tackle, applying the XAI methods they have learned to deepen their understanding.
