AI Tools for Actuaries
About This Project
This project aims to empower the actuarial profession with modern machine learning and AI tools. We provide comprehensive teaching materials that consist of lecture notes (technical document) building the theoretical foundation of this initiative. Each chapter of these lecture notes is supported by notebooks and slides which give teaching material, practical guidance and applied examples. Moreover, hands-on exercises in both R and Python are provided in additional notebooks.
Lecture Notes (Technical Document)
Notebooks, Slides and Code
- Chapter 0: Use Case: Actuarial Modeling
- Chapter 1: Introduction and Preliminaries
- Chapter 2: Regression Models
- Chapter 3: Generalized Linear Models
- Chapter 4: Interlude
-
Chapter 5: Feed-Forward Neural Networks
The Theory of FNNs
FNN One-Hot Encoding Example
FNN Embedding Example
CANN Example
LocalGLMnet Example
- Chapter 6: Regression Trees and Random Forests
- Chapter 7: Gradient Boosting Machines
-
Chapter 8: Deep Learning for Tensor and Unstructured Data
Tensors and Unstructured Data
Convolutional Neural Networks
Recurrent Neural Networks
Attention Layers and Transformers
Mortality Example: CNN, RNN, Transformer
Credibility Transformer
-
Chapter 9: Explainability
Explainability (without SHAP)
SHapley Additive exPlanations (SHAP)
-
Chapter 10: Unsupervised Learning
Auto-Encoder - Dimension Reduction
Clustering Methods
Visualization Methods
- Chapter 11: Deep Generative Models
- Chapter 12: Large Language Models
-
Chapter 13: Reinforcement Learning
Tabular Reinforcement Learning
Actor-Critic Reinforcement Learning
- Chapter 14: Data Visualization