Simulation: Data with Skewed Marginal Distributions and Gaussian Copula (Simulated)
Description
This simulated benchmark data follows the model with skewed normal distribution as the marginal distribution and Gaussian copula
as the dependence structure. It can be used to demonstrate the usage of methods and validation. DST is the Data
Simulated for Training purpose; DSV is the Data Simulated for Validation (prediction) purpose.
Usage
data(data.simulation) #load data sets: DST and DSV
DST # data simulated for training
DSV # data simulated for validation
Format
Each of DST and DSV is a list containing the following components:
-
obs - matrix of 100 observations (as row), with 80 timepoints for each
-
tp - vector of time points, with length 80
-
cp - vector of covariates for each subject, with length 100
-
pars - parameter list with
mean, logvar (matrices of 100 by 80) and shape, skew (vectors of length 80) -
corr - correlation matrix to determine the Gaussian corpula
In addition, each data set contains one specific component:
quantile- 3-dimensional array with dimension
c(5, 100, 80) for the true quantiles of the quantile levels
c(.50, .80, .90, .95, .99); only available for the training data set DST obs.full- fully observed observation matrix with diemsnion
c(100, 80), in the validation data set DSV; obs in DSV
contains missing values, and obs.full can be used to measure the performance of predication at those points where missing values are observed.