Categories
- ACTUARIAL DATA SCIENCE
- AFIR / ERM / RISK
- ASTIN / NON-LIFE
- BANKING / FINANCE
- DIVERSITY & INCLUSION
- EDUCATION
- HEALTH
- IACA / CONSULTING
- LIFE
- PENSIONS
- PROFESSIONALISM
- Thought Leadership
- MISC
This study proposes an enhanced sparse regularization technique for insurance ratemaking, focusing on automatic segmentation of rating classes. It introduces a group fused lasso regularization to group insurance rating factors into fewer categories, ensuring accurate and simple tariff referencing. The approach is demonstrated using motorcycle insurance data, showing effective grouping of adjacent categories into homogeneous clusters for expected claim frequency and severity.
0 Comments
There are no comments yet. Add a comment.