Extract Posterior Draws from Model Objects

S3 generic for extracting posterior draws from fitted model objects. Each method returns a standardized list with a draws matrix and optional parameter metadata, which can then be passed to der_compute().

Usage

# S3 method for class 'brmsfit'
extract_draws(x, ..., pars = NULL)

# S3 method for class 'CmdStanMCMC'
extract_draws(x, ..., pars = NULL)

# S3 method for class 'draws_matrix'
extract_draws(x, ...)

# S3 method for class 'draws_df'
extract_draws(x, ...)

# S3 method for class 'draws_array'
extract_draws(x, ...)

# S3 method for class 'draws_list'
extract_draws(x, ...)

# S3 method for class 'draws_rvars'
extract_draws(x, ...)

# S3 method for class 'stanreg'
extract_draws(x, ..., pars = NULL)

extract_draws(x, ...)

# S3 method for class 'matrix'
extract_draws(x, ...)

# Default S3 method
extract_draws(x, ...)

Arguments

x

A fitted model object (matrix, brmsfit, CmdStanMCMC, stanreg, draws_matrix, or draws_df).

...

Additional arguments passed to methods.

pars

Character vector of parameter names to extract from the stanreg object. If NULL (default), extracts all fixed effect and group-level parameters, excluding variance components and log-posterior.

Value

A list with components:

draws

Numeric matrix of posterior draws (S x d), where S is the number of posterior samples and d is the number of parameters.

param_info

(Optional) Data frame with parameter metadata including original names and mapped names.

Details

Methods are provided for matrix (identity), brmsfit, CmdStanMCMC, stanreg, draws_matrix, draws_df, draws_array, draws_list, and draws_rvars.

See also

extract_design() for extracting survey design information, der_compute() for computing DER from draws.

Other extraction: extract_design()

Examples

# Matrix method is the identity
m <- matrix(rnorm(100), nrow = 20)
result <- extract_draws(m)
identical(result$draws, m)
#> [1] TRUE