predict.BmaSamples: Predict method to extract point and interval predictions from...

In bfp: Bayesian Fractional Polynomials

Predict method to extract point and interval predictions from BmaSamples objects

Description

Predict new responses from a Bayesian model average over FP models, from which predictive samples have already been produced.

Usage

## S3 method for class 'BmaSamples'
predict(object, level=0.95, hpd=TRUE, ...)
## S3 method for class 'predict.BmaSamples'
print(x, ...)

Arguments

object	valid BmaSamples object
level	credible level for the credible intervals (default: 95%)
hpd	should emprical hpd intervals be used (default) or simple quantile-based?
...	unused
x	object of S3 class predict.BmaSamples

Details

This function summarizes the predictive samples saved in the BmaSamples object. Using these functions, one can obtain predictive credible intervals, as opposed to just using the function bmaPredict, which only gives the means of the predictive distributions.

Value

A list of class predict.BmaSamples, which has then a separate print method. The elements of the list are:

intervalType	which credible intervals have been computed (either “HPD” or “equitailed”)
level	the credible level
newdata	the covariate data for the predicted data points (just copied from object)
sampleSize	the sample size (just copied from object)
nModels	the number of models (just copied from object)
summaryMat	the summary matrix for the predictions, with median, mean, lower and upper credible interval borders.

Author(s)

Daniel Saban\'es Bov\'e

See Also

bmaPredict

Examples

## generate a BmaSamples object
set.seed(19)

x1 <- rnorm(n=15)
x2 <- rbinom(n=15, size=20, prob=0.5) 
x3 <- rexp(n=15)

y <- rt(n=15, df=2)

test <- BayesMfp(y ~ bfp (x2, max = 4) + uc (x1 + x3), nModels = 100,
                 method="exhaustive")

## predict new responses at (again random) covariates with BMA:
testBma <- BmaSamples(test,
                      newdata=data.frame(x1 = rnorm (15),
                                         x2 = rbinom (n=15, size=5, prob=0.2) + 1,
                                         x3 = rexp (15)))
predict(testBma)

bfp documentation built on Nov. 1, 2022, 1:05 a.m.