MSIMST (version 1.1)

reg_simulation3: The Function for the Simulation Study with the Variable Selection and Survey Weights

Description

This simulation study is designed to show the effectiveness of the grouped horseshoe prior for the variable selection and the WFPBB() function for adjusting survey weights.

Usage

reg_simulation3(
  N,
  ni_lambda,
  beta,
  beta_b,
  dsq,
  sigmasq,
  delta,
  nu,
  muz,
  rho,
  sigmasq_z,
  zeta0,
  zeta1
)

Value

A simulated dataset with the response variable y, the design matrix X and the survey weight survey_weight.

Arguments

N

The number of subjects.

ni_lambda

The mean of Poisson distribution.

beta

The covariates' coefficients. A 10 by 1 vector.

beta_b

The slope of PD response.

dsq

A part of covariance parameter.

sigmasq

A part of covariance parameter.

delta

The skewness parameter.

nu

The degree of freedom.

muz

The location parameter of the latent/selection variable.

rho

The correlation parameter of the latent/selection variable.

sigmasq_z

The variance parameter of the latent/selection variable.

zeta0

The intercept term inside the logistic function.

zeta1

The slope term inside the logistic function.

Details

More details of the design of this simulation study can be found in the vignette. Users can access the vignette by the command vignette(package = "MSIMST").

Examples

set.seed(100)
output_data <- reg_simulation3(N = 1000,
                               ni_lambda= 8,
                               beta = c(rep(1,6),rep(0,4)),
                               beta_b = 1.5,
                               dsq = 0.1,
                               sigmasq = 0.5,
                               delta = 0.6,
                               nu = 5.89,
                               muz = 0,
                               rho = 36.0,
                               sigmasq_z = 0.6,
                               zeta0 = -1.8,
                               zeta1 = 0.1)
y <- output_data$y
X <- output_data$X
survey_weight <- output_data$survey_weight