Each hub collects every post on a topic in one place — tutorials, benchmarks, library comparisons — alongside the open-source library that implements it.
Conformal Prediction for Insurance
Distribution-free prediction intervals with finite-sample coverage guarantees. No distributional assumptions, no model-specific variance estimation. Works on Tweedie GBMs, frequency-severity models, and reserve ranges.
Library: insurance-conformal
Causal Inference for Insurance Pricing
Double machine learning for deconfounding rating factors and estimating price elasticity. Causal forests for heterogeneous treatment effects. Synthetic DiD for rate change evaluation and FCA evidence packs.
Library: insurance-causal
Insurance Model Monitoring
Exposure-weighted PSI and CSI, A/E ratios with Poisson confidence intervals, Gini drift z-tests, and double-lift monitoring for deployed pricing models. Layered detection that fires months before your A/E ratio moves.
Library: insurance-monitoring
SHAP Relativities for Insurance
Multiplicative rating factor tables from CatBoost models in GLM exp(beta) format. Confidence intervals, exposure weighting, and direct export to Radar and Emblem. GBM-to-GLM distillation for rating engine deployment.
Library: shap-relativities
Insurance Fairness and Proxy Discrimination
Proxy discrimination auditing aligned to FCA EP25/2 and Consumer Duty. Optimal transport discrimination-free pricing. Fairness testing without sensitive attribute data. Structured documentation output.
Library: insurance-fairness