Helper function to generate simulated errors.

Generate simulated errors from the specified error function and passed arguments.

Usage

generate_errors(err = NULL, n, X, ...)

Arguments

err

Function from which to generate simulate error vector. Default is NULL which returns an error vector of all zeros.

n

Number of samples.

X

Data matrix or data frame. Used to determine n if n is missing.

...

Other arguments to pass to err() to generate the error vector.

Value

A vector of simulated errors with length n.

Details

The arguments n and X (if provided) are automatically passed to the function err under arguments of the same name. Note however that they may be unused arguments if err does not take in arguments named n and/or X as input.

Examples

# generate standard Gaussian error vector of length 150
errs <- generate_errors(err = rnorm, n = 150)
# or alternatively,
errs <- generate_errors(err = rnorm, X = iris)

# generate Gaussian error vector with mean 0 and sd 2
errs <- generate_errors(err = rnorm, n = 150, sd = 2)

# generate error vector of all 0s
errs <- generate_errors(err = NULL, n = 150)

# generate error vector from custom error function
err_fun <- function(n, rho) {
  # simulate correlated errors from a autoregressive-1 Gaussian process
  row1 <- rho^(0:(n - 1))
  Sigma <- stats::toeplitz(row1)
  return(MASS::mvrnorm(1, mu = rep(0, n), Sigma = Sigma))
} 
errs <- generate_errors(err = err_fun, n = 100, rho = 0.75)