WS-2 - From Glms to Explainable Machine Learning

In this interactive workshop, we will embark on an insightful trek through the evolution of actuarial predictive modeling. Using a real-life insurance dataset, we'll traverse from the foundational Generalized Linear Models (GLMs) to advanced techniques, including regularized GLMs and Gradient Boosting Machines (GBMs).

Participants will roll up their sleeves and delve into exploratory data analysis, data partitioning, iterative model selection, specification, hyperparameter optimization, and validation. Our emphasis? Drawing rich interpretations and gaining insights from the heart of the model outputs.

While we'll touch upon the intricate gears that drive these models, our compass is set firmly on comprehending the 'why' and 'how' behind each technique and understanding its tangible implications. As a highlight, we'll dive deep into cutting-edge explainability tools like SHAP and partial dependence plots, ensuring our models are not just high performers but also crystal-clear in their revelations.

Our workshop will be conducted exclusively in the R language. Attendees will receive all datasets and R code snippets in advance to ensure a smooth and engaging workshop experience.