augment.loo: loo tidiers

In jrnold/bayz: Miscellaneous Bayesian functions

Description

Methods for tidying the results from from WAIC and LOO cross-validation methods in the loo package.

Usage

augment.loo(x, data = NULL, ...)

glance.loo(x, ...)

tidy.loo(x, ...)

Arguments

x	an loo object
data	Data frame with the original data
...	Not used currently

Value

augment.loo returns a data frame, with one row per observation and the pointwise elpd, elpd standard error, and estimated number of observations.

glance.loo returns a one row data frame with the columns

elpd_loo,elpd_waic	Expected log pointwise predictive density
se_elpd_loo,se_elpd_waic	Standard error of the log pointwise predictive density
se_looic, see_waic	LOO information criterion (-2 * elpd_loo), or WAIC, i.e. converted to the deviance scale
p_loo, p_waic	Estimated effective number of parameters
se_p_loo, se_p_waic	Standard error of the estimated effective number of parameters
n	Number of observations
n_sims	Number of MCMC sims

A data frame with columns:

parameter	Parameter: elpd_loo or elpd_waic, p_loo or p_waic, or waic or looic
estimate	Parameter value
std.error	Standard error of the parameter

jrnold/bayz documentation built on May 5, 2019, 5:52 p.m.