data.generation {mvnimpute}

Data generation function

Description

Simulates multivariate normal data with missing and censored values. In this function, missing values will be generated first in the multivariate data, then censored values will be generated for the non-missing data.

Usage

data.generation(
  num_ind = 2000,
  mean_vec = rnorm(5),
  cov_mat = diag(5),
  miss_var = c(2, 3),
  miss_mech = "MCAR",
  miss_prob = c(0.2, 0.4),
  censor_var = 4,
  censor_type = "interval",
  censor_param = 0.1
)

Arguments

num_ind

number of subjects.

mean_vec

mean vectors.

cov_mat

covariance matrix.

miss_var

variables that have missing values.

miss_mech

missing mechanism. "MCAR" or "MAR". Default "MCAR".

miss_prob

missing data probability when missing data is MCAR.

censor_var

variables that have censored values.

censor_type

type of censoring. "interval", "right" or "left. Default "interval".

censor_param

rate parameter of the exponential distribution that the censoring times come from.

Value

A list containing the fully observed data, the observed data, the bounds information of the observed data and the data type indicator matrix.

Examples

### generate a multivariate normal dataset of 2000 sample size
### using the default arguments
data.generation()