pymc-modeling

Modern Bayesian modeling with PyMC v5+. Key defaults: nutpie sampler (2-5x faster), non-centered parameterization for hierarchical models, HSGP over exact GPs, coords/dims for readable InferenceData, and save-early workflow to prevent data loss from late crashes.

Modeling strategy: Build models iteratively — start simple, check prior predictions, fit and diagnose, check posterior predictions, expand one piece at a time. See references/workflow.md for the full workflow.

Notebook preference: Use marimo for interactive modeling unless the project already uses Jupyter.

Modellazione statistica bayesiana con PyMC v5+. Da utilizzare quando si creano modelli probabilistici, si specificano i valori a priori, si esegue l'inferenza MCMC, si diagnostica la convergenza o si confrontano modelli. Copre i backend PyMC, ArviZ, pymc-bart, pymc-extras, nutpie e JAX/NumPyro. Trigger su attività che coinvolgono: inferenza bayesiana, campionamento a posteriori, modelli gerarchici/multilivello, GLM, serie temporali, processi gaussiani, BART, modelli misti, controlli predittivi precedenti/posteriori, diagnostica MCMC, LOO-CV, WAIC, confronto di modelli o inferenza causale con do/observe. Fonte: pymc-labs/python-analytics-skills.