importance_sampling

An array, matrix, or vector of importance ratios on the log
scale (for PSIS-LOO these are negative log-likelihood values). See the
Methods (by class) section below for a detailed description of how
to specify the inputs for each method.

log_ratios

The importance sampling method to use. The following methods
are implemented:<ul>
<li><code><a rd-options="=psis" href="/link/%22psis%22?package=loo&version=2.3.1&to=%3Dpsis" data-mini-rdoc="=psis::"psis"">"psis"</a></code>: Pareto-Smoothed Importance Sampling (PSIS). Default method.</li>
<li><code><a rd-options="=tis" href="/link/%22tis%22?package=loo&version=2.3.1&to=%3Dtis" data-mini-rdoc="=tis::"tis"">"tis"</a></code>: Truncated Importance Sampling (TIS) with truncation at
<code>sqrt(S)</code>, where <code>S</code> is the number of posterior draws.</li>
<li><code><a rd-options="=sis" href="/link/%22sis%22?package=loo&version=2.3.1&to=%3Dsis" data-mini-rdoc="=sis::"sis"">"sis"</a></code>: Standard Importance Sampling (SIS).</li>
</ul>

method

Arguments passed on to the various methods.

A parent class for different importance sampling methods.

internal

Efficient approximate leave-one-out cross-validation (LOO)
for Bayesian models fit using Markov chain Monte Carlo, as
described in Vehtari, Gelman, and Gabry (2017)
<doi:10.1007/s11222-016-9696-4>.
The approximation uses Pareto smoothed importance sampling (PSIS),
a new procedure for regularizing importance weights.
As a byproduct of the calculations, we also obtain approximate
standard errors for estimated predictive errors and for the comparison
of predictive errors between models. The package also provides methods
for using stacking and other model weighting techniques to average
Bayesian predictive distributions.

Jonah Gabry

Efficient Leave-One-Out Cross-Validation and WAIC for Bayesian
Models

Aki Vehtari

Mans Magnusson

Yuling Yao

Paul-Christian B<c3><bc>rkner

Topi Paananen

Andrew Gelman

Ben Goodrich

Juho Piironen

importance_sampling function

The importance sampling method to use. The following methods
are implemented:<ul>
<li><code><a rd-options='=psis' href='"psis"'>"psis"</a></code>: Pareto-Smoothed Importance Sampling (PSIS). Default method.</li>
<li><code><a rd-options='=tis' href='"tis"'>"tis"</a></code>: Truncated Importance Sampling (TIS) with truncation at
<code>sqrt(S)</code>, where <code>S</code> is the number of posterior draws.</li>
<li><code><a rd-options='=sis' href='"sis"'>"sis"</a></code>: Standard Importance Sampling (SIS).</li>
</ul>

importance_sampling: A parent class for different importance sampling methods.

Description

Usage

Arguments