Predict from a Tensor Regression Fit

Usage

# S3 method for class 'spgtr'
predict(
  object,
  newX = NULL,
  newZ = NULL,
  type = c("response", "link", "class"),
  ...
)

Arguments

object: A fit from spgtr() or spgtr_cv().
newX: New subjects, in either form accepted by spgtr(). If omitted, predictions for the subjects used in fitting are returned.
newZ: New ordinary covariates, required when the fit used Z.
type: "response" (default) for the outcome scale – a probability for binomial(), an expected count for poisson(); "link" for the linear predictor; "class" for a 0/1 label thresholded at 0.5 (binomial fits only).
...: Unused.

Value

A numeric vector with one prediction per subject.

Examples

set.seed(3)
X <- lapply(1:80, function(i) matrix(rnorm(20), 5, 4))
y <- rbinom(80, 1, 0.5)
fit <- spgtr(X, y, u = c(1, 1))
predict(fit, X[1:5], type = "response")
#> [1] 0.6807172 0.3221624 0.6238999 0.6248340 0.3706757