Selected work
Actuarial
Health Insurance Pricing Model
Frequency-severity pricing with an interactive interface
An end-to-end actuarial pricing model implementing a classical frequency-severity framework and exposing the results through an interactive R Shiny interface.
POISSON × LOGNORMAL GLMModel output · synthetic data
Male, BMI 25, no children, before loading
Problem
Pricing a health portfolio requires separating how often claims occur from how large they are, then combining both components into an expected-cost and premium framework.
What I built
Claim frequency is modelled with a Poisson GLM and severity with a lognormal specification. Age, BMI, smoking status and dependants feed the risk classification, expected cost and premium calculation surfaced through a Shiny interface.
Methods
- Poisson GLM for claim frequency
- Lognormal model for claim severity
- Risk classification using member characteristics
- Expected-cost estimation combining frequency and severity
- Interactive risk-based premium calculation
Research context
- Built and tested using a synthetic portfolio.
Technology
- R
- GLMs
- Shiny