Streamlines the creation, sampling, and diagnostics of Bayesian statistical models using PyMC and ArviZ.
The PyMC skill enhances Claude's ability to assist with probabilistic programming and Bayesian inference. It provides specialized knowledge for building complex statistical models, including hierarchical and multilevel structures, while ensuring best practices like non-centered parameterization. By integrating MCMC sampling patterns, variational inference, and ArviZ-based diagnostics, this skill helps users perform robust uncertainty quantification, model comparison, and predictive checks directly within their development workflow.
主な機能
01Comprehensive model diagnostics including R-hat and ESS checks
020 GitHub stars
03Posterior and prior predictive checks for model validation
04Optimized MCMC sampling configurations using the NUTS sampler
05Advanced uncertainty quantification and credible interval estimation
06Expert construction of linear, logistic, and hierarchical Bayesian models
ユースケース
01Developing hierarchical models to account for group-level variations in data
02Quantifying prediction uncertainty for risk assessment and decision making
03Comparing multiple statistical models using LOO and WAIC metrics
