predict.bkmrfit {bkmrhat}R Documentation

Posterior mean/sd predictions

Description

Provides observation level predictions based on the posterior mean, or, alternatively, yields the posterior standard deviations of predictions for an observation. This function is useful for interfacing with ensemble machine learning packages such as SuperLearner, which utilize only point estimates.

Usage

## S3 method for class 'bkmrfit'
predict(object, ptype = c("mean", "sd.fit"), ...)

Arguments

object

fitted object of class inheriting from "bkmrfit".

ptype

"mean" or "sd.fit", where "mean" yields posterior mean prediction for every observation in the data, and "sd.fit" yields the posterior standard deviation for every observation in the data.

...

arguments to SamplePred

Value

vector of predictions the same length as the outcome in the bkmrfit object

Examples

# following example from https://jenfb.github.io/bkmr/overview.html

library(bkmr)
set.seed(111)
dat <- bkmr::SimData(n = 50, M = 4)
y <- dat$y
Z <- dat$Z
X <- dat$X
set.seed(111)
fitkm <- kmbayes(y = y, Z = Z, X = X, iter = 200, verbose = FALSE,
  varsel = TRUE)
postmean = predict(fitkm)
postmean2 = predict(fitkm, Znew=Z/2)
# mean difference in posterior means
mean(postmean-postmean2)


[Package bkmrhat version 1.1.1 Index]