About This Event
GBMs present a significant opportunity for improved predictive power over GLMs. However, insurance practitioners know they need to balance other priorities, such as transparency and stability – areas that can be weaknesses for GBMs. After a brief introduction to GBMs, we will discuss specific issues related to their use for insurance applications. We will then demonstrate different approaches to address these issues, focusing on answering questions such as:
- For what types of problems does it make sense to use GBMs instead of GLMs?
- How can I understand the predictions a GBM makes?
- How can I ensure my GBM is sufficiently stable?
