predict.bfa: Posterior predictive and univariate conditional posterior...

In bfa: Bayesian Factor Analysis

Description

Posterior predictive and univariate conditional posterior predictive distributions, currently implemented only for copula models. If resp.var is not NA, returns an estimate of the conditional cdf at every observed data point for each MCMC iterate. If resp.var is NA, returns draws from the joint posterior predictive.

Usage

  ## S3 method for class 'bfa'
 predict(object, resp.var = NA, cond.vars =
  NA, numeric.as.factor = TRUE, ...)

Arguments

object	bfa model object
resp.var	Either a character vector (length 1) with name of the response variable for conditional, or NA for draws from the joint posterior predictive.
cond.vars	Conditioning variables; either a list like list(X1=val1, X2=val2) with X1, X2 variables in the original data frame, or a P length vector with either the conditioning value or NA (for marginalized variables). Ignored if resp.var is NA
numeric.as.factor	Treat numeric variables as ordinal when conditioning
...	Ignored

Value

A matrix where each row is either a sample of the conditional posterior predictive cdf at each datapoint, or a single sample from the joint posterior predictive.

bfa documentation built on May 2, 2019, 4:55 p.m.