predict.shrinkMVGPR: Generate Predictions

In shrinkGPR: Scalable Gaussian Process Regression with Hierarchical Shrinkage Priors

Generate Predictions

Description

predict.shrinkMVGPR generates posterior predictive samples from a fitted shrinkMVGPR model at specified covariates.

Usage

## S3 method for class 'shrinkMVGPR'
predict(object, newdata, nsamp = 100, ...)

Arguments

object	A shrinkMVGPR or shrinkMVTPR object representing the fitted multivariate process regression model.
newdata	Optional data frame containing the covariates for the prediction points. If missing, the training data is used.
nsamp	Positive integer specifying the number of posterior samples to generate. Default is 100.
...	Currently ignored.

Details

This function generates predictions by sampling from the posterior predictive distribution.

Value

A 3-dimensional array of dimensions nsamp x N_new x M containing posterior predictive samples for each covariate combination in newdata.

Examples

if (torch::torch_is_installed()) {
  # Simulate data
  set.seed(123)
  torch::torch_manual_seed(123)
  n <- 100
  x <- matrix(runif(n * 2), n, 2)
  y1 <- sin(2 * pi * x[, 1])
  y2 <- cos(2 * pi * x[, 2])
  y <- cbind(y1, y2) + matrix(rnorm(n * 2, sd = 0.1), n, 2)
  data <- data.frame(y1 = y1, y2 = y2, x1 = x[, 1], x2 = x[, 2])

  # Fit MVGPR model
  res <- shrinkMVGPR(cbind(y1, y2) ~ x1 + x2, data = data)
  # Example usage for in-sample prediction
  preds <- predict(res)

  # Example usage for out-of-sample prediction
  newdata <- data.frame(x1 = runif(10), x2 = runif(10))
  preds <- predict(res, newdata = newdata)
  }

shrinkGPR documentation built on March 30, 2026, 5:06 p.m.