R/data.R

In BrainCon: Inference the Partial Correlations Based on Time Series Data

#' Simulation time series data for individual
#'
#' A dataset containing values of 10 interested
#' variables over 50 periods.
#'
#' @examples
#' ## Generated by the following R codes
#' set.seed(1000)
#' n = 50; p = 10
#' Precision = diag(rep(2, p))        # generate precision matrix
#' for (i in 1 : (p - 1)){
#'   temp = ifelse(i > 2 * p / 3, 0.4, 1)
#'   Precision[i, i + 1] = temp
#'   Precision[i + 1, i] = temp
#' }
#' # R=-cov2cor(Precision) + diag(rep(2, p)) # real partial correlation matrix
#' Sigma = solve(Precision)           # generate covariance matrix
#' rho = 0.5
#' y = matrix(0, n, p)                # generate observed time series data
#' Epsilon = MASS::mvrnorm(n, rep(0, p), Sigma)
#' y[1, ] = Epsilon[1, ]
#' for (i in 2 : n){
#'   y[i, ] = rho * y[i - 1, ] + sqrt(1 - rho^2) * Epsilon[i, ]
#' }
#' indsim = y
"indsim"

#' Simulation time series data for population A
#'
#' A dataset containing values of 10 interested
#' variables of 20 subjects over 50 periods.
#' @seealso \code{\link{popsimB}}.
#' @examples
#' ## Generated by the following R codes
#' set.seed(1234)
#' n = 50; p = 10; m1 = 20; m2 = 10
#' Precision1 = Precision2 = diag(rep(1, p))    # generate Precision matrix for population
#' for (i in 1 : (p - 1)){
#'   temp1 = ifelse(i > 2 * p / 3, -0.2, 0.4)
#'   temp2 = ifelse(i < p / 3, 0.4, -0.2)
#'   Precision1[i, i + 1] = Precision1[i + 1, i] = temp1
#'   Precision2[i, i + 1] = Precision2[i + 1, i] = temp2
#' }
#' # R1=-cov2cor(Precision1) + diag(rep(2, p))  # real partial correlation matrix
#' # R2=-cov2cor(Precision2) + diag(rep(2, p))
#' Index = matrix(0, p, p)                      # generate covariance matrix for each subject
#' for (i in 1 : p){
#'   for (j in 1 : p){
#'     if (i != j & abs(i - j) <= 3) Index[i, j] = 1
#'   }
#' }
#' SigmaAll1 = array(dim = c(p, p, m1))
#' SigmaAll2 = array(dim = c(p, p, m2))
#' for (sub in 1 : m1){
#'   RE = matrix(rnorm(p^2, 0, sqrt(2) * 0.05), p, p) * Index
#'   RE1 = (RE + t(RE)) / 2
#'   PrecisionInd = Precision1 + RE1
#'   SigmaAll1[, , sub] = solve(PrecisionInd)
#' }
#' for (sub in 1 : m2){
#'   RE = matrix(rnorm(p^2, 0, sqrt(2) * 0.15), p, p) * Index
#'   RE1 = (RE + t(RE)) / 2
#'   PrecisionInd = Precision2 + RE1
#'   SigmaAll2[, , sub] = solve(PrecisionInd)
#' }
#' rho = 0.3                                    # generate observed time series data
#' y1 = array(dim = c(n, p, m1))
#' y2 = array(dim = c(n, p, m2))
#' for (sub in 1 : m1){
#'   SigmaInd1 = SigmaAll1[, , sub]
#'   ytemp = matrix(0, n, p)
#'   Epsilon = MASS::mvrnorm(n, rep(0, p), SigmaInd1)
#'   ytemp[1, ] = Epsilon[1, ]
#'   for (i in 2 : n){
#'     ytemp[i, ] = rho * ytemp[i - 1, ] + sqrt(1 - rho^2) * Epsilon[i, ]
#'   }
#'   y1[, , sub] = ytemp
#' }
#' for (sub in 1 : m2){
#'   SigmaInd2 = SigmaAll2[, , sub]
#'   Xtemp = matrix(0, n, p)
#'   Epsilon = MASS::mvrnorm(n, rep(0, p), SigmaInd2)
#'   ytemp[1, ] = Epsilon[1, ]
#'   for (i in 2 : n){
#'     ytemp[i, ] = rho * ytemp[i - 1, ] + sqrt(1 - rho^2) * Epsilon[i, ]
#'   }
#'   y2[, , sub] = ytemp
#' }
#' popsimA = y1
#' popsimB = y2
"popsimA"

#' Simulation time series data for population B
#'
#' A dataset containing values of 10 interested
#' variables of 10 subjects over 50 periods.
#'
#' @seealso \code{\link{popsimA}}.
"popsimB"