getInclusionProbs: Extract posterior inclusion probabilities from the models...

In hypergsplines: Bayesian model selection with penalised splines and hyper-g prior

Description

Extract posterior inclusion probabilities from the models table

Usage

  getInclusionProbs(models, modelData,
    logMargLiks = models$logMargLik,
    logPriors = models$logPrior)

Arguments

models	the models table (a data frame), the result from exhaustive
modelData	data used for model estimation
logMargLiks	vector of log marginal likelihoods (defaults to the logMargLik column of models)
logPriors	vector of log prior model probabilities (defaults to the logPrior column of models)

Value

a nice matrix with the probabilities for exclusion, linear and non-linear inclusion of the covariates

Author(s)

Daniel Sabanes Bove [e-mail]

Examples

## get some data
attach(longley)

## get model data
md <- modelData(y=Employed,
                X=cbind(GNP, Armed.Forces))

## get a list of all possible models with this data
models <- exhaustive(md,
                     modelPrior="exponential")$models

## attach log prior probabilities
models$logPrior2 <- apply(models[, 1:2],
                          1L,
                          getLogModelPrior,
                          type="exponential",
                          modelData=md)

stopifnot(all.equal(models$logPrior,
                    models$logPrior2))

## then we can compute the inclusion probabilities
getInclusionProbs(models=models,
                  modelData=md)

hypergsplines documentation built on May 2, 2019, 6:14 p.m.