Builds and validates complex Bayesian models for probabilistic inference using PyMC and ArviZ.
The PyMC Bayesian Modeling skill equips Claude with specialized expertise in probabilistic programming and statistical inference. It facilitates a robust, end-to-end Bayesian workflow, encompassing data standardization, weakly informative prior selection, MCMC sampling via the NUTS algorithm, and rigorous model diagnostics. This skill is particularly valuable for data scientists and researchers needing to implement hierarchical models, quantify uncertainty in predictions, or handle missing data through principled probabilistic methods, ensuring all models adhere to modern PyMC 5.x+ best practices.
主な機能
01Model comparison workflows using LOO and WAIC information criteria
02Automated model diagnostics for R-hat, ESS, and divergences
03Prior and posterior predictive check (PPC) integration
04Standardized MCMC sampling using NUTS and variational inference (ADVI)
05Hierarchical modeling templates with non-centered parameterization
061 GitHub stars
ユースケース
01Implementing uncertainty-aware regression for scientific or financial forecasting
02Building hierarchical/multilevel models for grouped data structures
03Performing robust model selection and averaging for complex datasets
