Apply Bayesian methods to statistical inference

Perform Bayesian statistical inference with prior specification, MCMC, and posterior computation

Incorporating prior domain knowledge into inference and updating beliefs with small sample sizes where frequentist confidence intervals fail

• · prior distribution
• · likelihood function
• · observed data
• · MCMC sampler config (optional)

• · posterior distribution
• · credible intervals
• · Bayes factors for model comparison

• · pymc
• · arviz
• · numpy
• · scipy

Data is numerical; prior must be parameterized; likelihood computable for data; MCMC requires convergence diagnostics

Poor prior leads to overconfident posteriors; MCMC divergence if step size too large; improper priors cause non-normalizable posteriors

• · Contrasts clearly with frequentist paradigm
• · Covers conjugate families for closed-form posteriors
• · Hierarchical models enable borrowing strength