makeExampleData: Simulate example data from the graper model

In graper: Adaptive penalization in high-dimensional regression and classification with external covariates using variational Bayes

Description

Simulate data from the graper model with groups of equal size and pre-specified parameters gamma, pi and tau.

Usage

makeExampleData(n = 100, p = 200, g = 4, gammas = c(0.1, 1, 10,
  100), pis = c(0.5, 0.5, 0.5, 0.5), tau = 1, rho = 0,
  response = "gaussian", intercept = 0)

Arguments

n	number of samples
p	number of features
g	number of groups
gammas	vector of length g, specifying the slab precision of the prior on beta per group
pis	vector of length g, specifying the probability of s to be 1 (slab)
tau	noise precision
rho	correlation of design matrix (Toeplitz structure)
response	"gaussian" for continuous response from a linear regression model, "bernoulli" for a binary response from a logistic regression model.
intercept	model intercept (default: 0)

Value

list containing the design matrix X, the response y, the feature annotation to groups annot as well as the different parameters in the Bayesian model and the correlation strength rho

Examples

dat <- makeExampleData()

graper documentation built on Nov. 8, 2020, 5:45 p.m.