OpenSkillIndex← back to search
claude-codescience

Babysitter pymc-bayesian-modeler

PyMC probabilistic programming skill for hierarchical Bayesian models in physics data analysis

install
source · Clone the upstream repo
git clone https://github.com/a5c-ai/babysitter
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/a5c-ai/babysitter "$T" && mkdir -p ~/.claude/skills && cp -r "$T/library/specializations/domains/science/physics/skills/pymc-bayesian-modeler" ~/.claude/skills/a5c-ai-babysitter-pymc-bayesian-modeler && rm -rf "$T"
manifest: library/specializations/domains/science/physics/skills/pymc-bayesian-modeler/SKILL.md
tags
#bayesian-modeling#hierarchical-models#physics-analysis#probabilistic-programming#data-analysis
source content

PyMC Bayesian Modeler

Purpose

Provides expert guidance on PyMC for Bayesian modeling in physics, including hierarchical models and advanced inference methods.

Capabilities

  • Probabilistic model construction
  • NUTS/HMC sampling
  • Variational inference
  • Gaussian processes
  • Model comparison (WAIC, LOO)
  • Prior predictive checks

Usage Guidelines

  1. Model Building: Construct probabilistic models
  2. Priors: Specify informative or weakly informative priors
  3. Sampling: Use NUTS for efficient sampling
  4. Diagnostics: Check convergence with trace plots and r-hat
  5. Comparison: Compare models with information criteria

Tools/Libraries

  • PyMC
  • arviz
  • Theano/JAX