$ ⌘K
// constructs / weakly_informative_prior

Weakly Informative Prior

id: weakly_informative_prior

A prior designed to rule out implausible parameter values without strongly favoring any specific value — typically a wide but proper distribution (e.g. Normal(0, 2.5) on standardized coefficients, half-Cauchy on variance). Provides regularization and numerical stability without imposing strong subjective beliefs. Recommended default for applied Bayesian workflow.

// usage

Used in 1 pax

↑ positive relationship 4 findings