Predict.matrix.unc_pcre.random.effect {MJMbamlss}

mgcv-style constructor for prediction of PC-basis functional random effects

Description

This predictor function uses time-information saved in the object. This is handled within the bamlss-transform function, so this function is not exported.

Usage

## S3 method for class 'unc_pcre.random.effect'
Predict.matrix(object, data)

Arguments

object

a smooth specification object, see smooth.construct

data

see smooth.construct

Value

design matrix for PC-based functional random effects

Author(s)

Alexander Volkmann, adapted from 'Predict.matrix.pcre.random.effect by F. Scheipl (adapted from 'Predict.matrix.random.effect' by S.N. Wood).

Examples

data(pbc_subset)
mfpca <- preproc_MFPCA(pbc_subset, uni_mean = paste0(
  "logy ~ 1 + sex + drug + s(obstime, k = 5, bs = 'ps') + ",
  "s(age, k = 5, bs = 'ps')"),
  pve_uni = 0.99, nbasis = 5, weights = TRUE, save_uniFPCA = TRUE)
pbc_subset <- attach_wfpc(mfpca, pbc_subset, n = 2)
mfpca_list <- list(
  list(functions = funData::extractObs(mfpca$functions, 1),
       values = mfpca$values[1]),
  list(functions = funData::extractObs(mfpca$functions, 2),
       values = mfpca$values[2]))
sm <- smoothCon(s(id, fpc.1, bs = "unc_pcre",
      xt = list("mfpc" = mfpca_list[[1]], scale = "FALSE")), pbc_subset)[[1]]
sm$timevar <- "obstime"
sm$term <- c(sm$term, "obstime")
pm <- PredictMat(sm, pbc_subset, n = 4*nrow(pbc_subset))