gfilmmPredictive: Generalized fiducial predictive distributions

In gfilmm: Generalized Fiducial Inference for Normal Linear Mixed Models

Generalized fiducial predictive distributions

Description

Simulations of the generalized fiducial predictive distributions.

Usage

gfilmmPredictive(gfi, newdata)

Arguments

gfi	a gfilmm object
newdata	dataframe in which to look for variables with which to predict, or NULL if the model is an intercept-only model without random effect

Value

A list with two fields: FPD, a dataframe containing the simulations, and WEIGHT, their weight. This is a gfilmm object.

Note

Actually the levels of the random effects given in newdata can be different from the original levels. For instance, in the example provided below, we enter block = c("4","6"), but we could also enter block = c("A","B"), even though "A" and "B" are not some levels of the block factor. Both options only mean that the two observations to predict are in two different blocks.

Examples

gfi <- gfilmm(
  ~ cbind(yield-0.1, yield+0.1), ~ N, ~ block, npk, 2000, nthreads = 2
)
fpd <- gfilmmPredictive(gfi, data.frame(N = c("0","1"), block = c("4","6")))
gfiSummary(fpd)

gfilmm documentation built on July 12, 2022, 1:05 a.m.