Overview
The nsEVDx library provides comprehensive tools for modeling extreme value distributions under both stationary and non-stationary conditions. Designed for hydrologists, climate scientists, and engineers working with extreme events, it implements both frequentist and Bayesian inference frameworks with flexible covariate modeling capabilities.
Core Features
Complete support for Generalized Extreme Value and Generalized Pareto distributions.
Time-varying parameters with arbitrary covariates for changing climate conditions.
Dual inference frameworks featuring robust MCMC sampling and MLE estimation.
High-performance MALA, HMC, and Metropolis-Hastings samplers for complex models.
Comprehensive convergence tracking and goodness-of-fit metrics for reliable results.
Built on the core scientific stack: numpy, scipy, matplotlib, and seaborn.
Key Applications
Analysis of extreme precipitation, temperature, and drought events under changing climate conditions.
Non-stationary flood frequency modeling with time-varying design quantiles for infrastructure planning.
Financial and environmental risk modeling with time-dependent extreme value parameters.
Design value estimation for structures under non-stationary extreme loads.