Dakota Reference Manual  Version 6.15
Explore and Predict with Confidence
 All Pages
model_evidence


Calculate model evidence (marginal likelihood of model) when using Bayesian methods

Specification

Alias: none

Argument(s): none

Child Keywords:

Required/Optional Description of Group Dakota Keyword Dakota Keyword Description
Optional mc_approx

Calculate model evidence using a Monte Carlo sampling approach

Optional evidence_samples

The number of samples used in Monte Carlo approximation of the model evidence.

Optional laplace_approx

Calculate model evidence using the Laplace approximation

Description

Model evidence is used in Bayesian model selection and model averaging. It is defined as the probability of the data given the model, and is calculated by averaging the likelihood of the model parameters over all values of the model parameters according to their prior distributions. In Dakota, one must calculate the model evidence separately for each model and perform the normalization to obtain the posterior model plausibility for each model.

Default Behavior

When specifying model_evidence, there are two methods of calculating it. One or both may be specified. They include the Monte Carlo approximation, given by mc_approx and the Laplace approximation, given by laplace_approx. mc_approx is the default approach.

Expected Output Currently, the model evidence will be printed in the screen output with prefacing text indicating if it is calculated by Monte Carlo sampling or the Laplace approximation.

Usage Tips