acturial

SQL analysis of car insurance claims to identify high-risk policyholder groups based on demographic, behavioural and vehicle-related attributes. This project calculates policyholder share, claim share, claim frequency and relative risk ratio to highlight groups with heightened claim behaviour.

Predict medical insurance charges using GLM (Gamma/log) and Gradient Boosting; includes calibration, permutation importance, and FastAPI demo.

Tweedie regression pipeline predicting Pure Premium (Expected Annual Liability) across 677k+ French Motor TPL insurance policies, with an interactive Streamlit dashboard.