A Python library to model stationary and non-stationary extreme value distributions (GEV & GPD)

nsEVDx

A Python library to model stationary and non-stationary extreme value distributions (GEV & GPD).

• PyPI
• GitHub

Key Features

• Stationary and non-stationary GEV / GPD models
• Arbitrary covariates in location, scale, and shape parameters
• Supports both Bayesian and Frequentist approaches
• Transparent, fully customizable MCMC engine implemented in NumPy
• Advanced samplers: Metropolis-Hastings, MALA, and HMC
• Minimal dependencies (NumPy, SciPy, matplotlib)

• Diagnostics: trace plots, acceptance rates, and Bayesian metrics

• The library is pip-installable and comes with Jupyter notebooks examples and documentation. Suggestions, issues, and contributions via GitHub are welcomed.