whitneysdata/insurance-claims-modelling: Insurance Claims Modelling - final version v3.0.0
This release contains the final version of my Insurance Claims Frequency–Severity Modelling project, built using the freMTPL2 French Motor Third Party Liability Insurance dataset (678,013 policies). The pipeline integrates statistical and machine learning methods to model claim frequency and severity. Frequency is estimated using Poisson and Negative Binomial regression, while severity is modelled with Gamma and Lognormal regression. To complement these approaches, Random Forest and XGBoost models are trained and benchmarked against traditional GLMs using metrics such as RMSE, MAE, and R‑squared. The combined predictions yield pure premium es
Resource information
- Format
- zip
- Original repository
- 20531814
- DOI
- 10.5281/zenodo.20531814
- Credit basis
- Creators in Zenodo record metadata
- Access
- Open files available from the original Zenodo deposit
- Last checked
- Nov 5, 2026
- License
- cc-by-4.0
- Last record update
- Nov 5, 2026
About this resource
This release contains the final version of my Insurance Claims Frequency–Severity Modelling project, built using the freMTPL2 French Motor Third Party Liability Insurance dataset (678,013 policies). The pipeline integrates statistical and machine learning methods to model claim frequency and severity. Frequency is estimated using Poisson and Negative Binomial regression, while severity is modelled with Gamma and Lognormal regression. To complement these approaches, Random Forest and XGBoost models are trained and benchmarked against traditional GLMs using metrics such as RMSE, MAE, and R‑squared. The combined predictions yield pure premium estimates, representing expected claim cost per unit of exposure. This release includes reproducible Python code, model outputs, and documentation, making it suitable for citation, replication, and further actuarial or data science research.