Creates a prior distribution specification for use in Bayesian analyses.
Usage
create_prior_specification(
distribution,
parameters,
description = "",
domain = "",
reference = "",
metadata = list()
)Arguments
- distribution
Character string specifying the distribution: "normal", "beta", "gamma", "half_cauchy", "half_normal", "exponential", "uniform", "cauchy", "t", "log_normal", "inverse_gamma", "laplace", "student_t"
- parameters
List of distribution parameters (see details)
- description
Character string describing the prior
- domain
Character string indicating the parameter domain
- reference
Character string with reference/citation
- metadata
List of additional metadata
Details
Distribution parameters:
- normal
mean, sd
- beta
shape1, shape2
- gamma
shape, rate (or scale)
- half_cauchy
location, scale
- half_normal
sd
- exponential
rate
- uniform
min, max
- cauchy
location, scale
- t
df, location, scale
- log_normal
meanlog, sdlog
- inverse_gamma
shape, scale
- laplace
location, scale
- student_t
df, location, scale
Examples
if (FALSE) { # \dontrun{
# Vague normal prior
prior1 <- create_prior_specification(
distribution = "normal",
parameters = list(mean = 0, sd = 10),
description = "Vague prior for overall effect",
domain = "overall_effect"
)
# Informative beta prior
prior2 <- create_prior_specification(
distribution = "beta",
parameters = list(shape1 = 2, shape2 = 2),
description = "Moderately informative prior for probability",
domain = "probability"
)
# Half-Cauchy prior for variance component
prior3 <- create_prior_specification(
distribution = "half_cauchy",
parameters = list(location = 0, scale = 0.5),
description = "Default prior for heterogeneity",
domain = "heterogeneity"
)
} # }