Goal
Driver age may relate to claim frequency, but claim counts need a model that respects count-data behavior.
portfolio project / statistical modeling
Auto Claims Frequency Modeling
Modeled synthetic auto insurance claim frequency by age group using a reproducible Python pipeline, statistical testing, and count-data regression.
Build
Created a Python workflow that generates synthetic data, engineers features, compares age groups, and fits Poisson and Negative Binomial models.
Result
Saved refreshed figures, summary tables, model output, and a public-ready README for a clean portfolio repo.
build notes
Choices and small problems
- Used synthetic claim data so the modeling flow can be shared without real insurance records.
- Compared Poisson and Negative Binomial models because claim counts can be overdispersed.
- Kept the age-group finding readable while noting that age alone is not enough for real pricing.
- Saved model tables and figures so the project can be checked without rerunning everything first.
visual notes
Actual claim graphs
These figures come from the auto-claims project output files.
Graph selector
Now viewing: Claim rate