extract.samples: Collect posterior or prior samples from a map or map2stan...

In rmcelreath/rethinking: Statistical Rethinking book package

Collect posterior or prior samples from a map or map2stan model

Description

Extracts or draw samples from fit models.

Usage

extract.samples( object , n=10000 , pars , ... )
extract.prior( object , n=1000 , pars , ... )

Arguments

object	Fit model to extract samples from
n	Number of samples to simulate
pars	Character vector of parameters to return
...	Other parameters to pass to descendent functions (when defined)

Details

Use extract.samples and extract.prior to return lists of samples for posterior and prior distributions, respectively. Methods for extract.samples are provided for map, map2stan, and stanfit objects. Methods for extract.prior are provided for map and map2stan objects.

For map2stan and stanfit models, extract.samples returns cleaned samples already contained in the object. These samples are cleaned of dimension attributes and the lp__, dev, and log_lik traces that are used internally. For map and other types, it uses the variance-covariance matrix and coefficients to define a multivariate Gaussian posterior to draw n samples from.

extract.prior must simulate draws using the model definition. It attempts to return samples in the same list structure as posterior samples. This makes prior-posterior comparison easier. See examples below.

Value

A named list (for map2stan and stanfit) or data.frame containing samples for each parameter in the posterior/prior distribution.

Author(s)

Richard McElreath

See Also

mvrnorm

Examples

data(chimpanzees)

d <- list( 
    pulled_left = chimpanzees$pulled_left ,
    prosoc_left = chimpanzees$prosoc_left ,
    condition = chimpanzees$condition ,
    actor = as.integer( chimpanzees$actor ) ,
    blockid = as.integer( chimpanzees$block )
)

m <- map(
    alist(
        pulled_left ~ dbinom(1,theta),
        logit(theta) <- a + aj[actor] + bp*prosoc_left + bpc*condition*prosoc_left,
        aj[actor] ~ dnorm( 0 , 1 ),
        a ~ dnorm(0,2),
        bp ~ dnorm(0,1),
        bpc ~ dnorm(0,1)
    ) ,
    data=d )

prior <- extract.prior(m,n=1e4)
post <- extract.samples(m)
ps <- par("bty")
par(bty="n")
plot( precis(prior,2) , col.ci="gray" , xlim=c(-3,3.5) , bty="n" )
plot( precis(post,2) , add=TRUE , pch=16 )
par(bty=ps)

rmcelreath/rethinking documentation built on Sept. 18, 2023, 2:01 p.m.