R/CNE.R

In sunbisunbi/CNE: Censored Network Estimation

#' @export
CNE = function(survdata, method = c('CNE', 'dabrowska', 'linying', 'corcoef','naive.X'), l=10000, cv=F, GEN.BOUND=T){
  method = match.arg(method)
  N.sample = nrow(survdata)
  dimension = ncol(survdata)/2
  if(method=='CNE'){
    Den = opt_(survdata, dimension = 2, repeats = TRUE);
    estimated.T = Estimate_T_MonteCalro(survdata, Den, GEN.BOUND = GEN.BOUND);
    estimated.cov = cov(estimated.T);
    estimated.edge = rho_glasso(estimated.cov, N.sample, l=l);
    if(cv){
      cv.rho = CVglasso(estimated.T)
      res = list(cv.rho = cv.rho[[3]][2], Edges = estimated.edge)
      return(res)
    }
  } else if(method=='dabrowska'){
    Den = probability_estimation(survdata, method = "dabrowska", repeats=TRUE);
    estimated.T = Estimate_T_MonteCalro(survdata, Den, GEN.BOUND = GEN.BOUND);
    estimated.cov = cov(estimated.T);
    estimated.edge = rho_glasso(estimated.cov, N.sample, l=l);
  } else if(method=='linying'){
    Den = probability_estimation(survdata, method = "linying", repeats=TRUE);
    estimated.T = Estimate_T_MonteCalro(survdata, Den, GEN.BOUND = GEN.BOUND);
    estimated.cov = cov(estimated.T);
    estimated.edge = rho_glasso(estimated.cov, N.sample, l=l);
  } else if(method=='corcoef'){
    estimated.edge = naive.cor(survdata, NUM=20)
  } else if(method=='naive.X'){
    X.cov = cov(survdata[,1:dimension*2-1])
    estimated.edge = rho_glasso(X.cov, N.sample, l=l);
  }
  colnames(estimated.edge) = c('Node1', 'Node2', 'rho')
  return(estimated.edge)
}