Builds, fits, and validates sophisticated Bayesian models using PyMC's probabilistic programming interface.
The PyMC Bayesian Modeling skill equips Claude with specialized knowledge for performing advanced probabilistic programming and inference in Python. It provides structured workflows for the entire Bayesian lifecycle—from data standardization and prior predictive checks to MCMC sampling with the NUTS sampler and variational inference. By implementing best practices like non-centered parameterization for hierarchical models and rigorous diagnostic checks (R-hat, ESS, divergences), this skill ensures your statistical models are both robust and computationally efficient. It is ideal for data scientists and researchers needing principled uncertainty quantification and model comparison using LOO or WAIC.
Características Principales
010 GitHub stars
02Automated sampling diagnostics for R-hat, Effective Sample Size (ESS), and divergences
03Comprehensive Bayesian workflows including prior and posterior predictive checks
04Pre-defined patterns for linear, logistic, hierarchical, and time-series models
05Guidance on non-centered parameterization to resolve complex sampling issues
06Advanced model comparison using ArviZ, LOO, and WAIC information criteria
Casos de Uso
01Performing robust uncertainty quantification in scientific or financial forecasting
02Developing hierarchical/multilevel models for grouped data structures
03Diagnosing and resolving NUTS sampling failures in complex probabilistic programs
