sim.gauss: Simulate data for benchmarking Gaussian regression models.

In abnormally-distributed/cvreg: Cross Validation and Robust Estimation Utilities

Description

Simulate data for benchmarking Gaussian regression models.

Usage

sim.gauss(
  n = 100,
  p = 25,
  rho = 0.5,
  coefs = c(1.5, 4, -2, -4, 1, 2, -2.5),
  snr = 2,
  x.df = Inf,
  scale = TRUE,
  cormat = NULL,
  seed = 100
)

Arguments

n	Number of observations.
p	Number of predictors.
rho	Correlation for generating correlated variables.
coefs	Vector of non-zero coefficients
snr	Signal to noise ratio (SNR). Defaults to 2. SNR is defined as \frac{Var(E(y | X))}{Var(Y - E(y | X))} = \frac{Var(f(X))}{Var(\varepsilon)} = \frac{Var(X^T β)}{Var(\varepsilon)} = \frac{Var(β^T Σ β)}{σ^2}.
x.df	The degrees of freedom for the multivariate distribution of covariates. Defaults to Inf (Gaussian).
scale	should the data be scaled? Defaults to TRUE.
seed	Random seed for reproducibility.

Value

a data frame with an attribute "true.betas" that contains the true coefficients. If scale = TRUE, the coefficients are scaled to match.

Author(s)

Brandon Vaughan

Examples

dat <- sim.gauss(
  n = 120, p = 200, rho = 0.26,
  coefs = c(runif(25, -4, -1), runif(25, 1, 4)), snr = 2,
  seed = 100)

abnormally-distributed/cvreg documentation built on May 3, 2020, 3:45 p.m.