R/bootnet.R

In skynet: Generates Networks from BTS Data

#' Network bootstrapping
#'
#' Bootstraps a network and returns output containing three network statistics:
#' Average Path Length, Transitivity, Mean Betweenness.
#'
#'
#' @param g iGraph graph or skynet object.
#' @param n Number of bootstraps to run. (500 default)
#' @param left_ci Confidence interval left limit. (0.005 default)
#' @param right_ci Confidence interval left limit (0.995 default)
#'
#' @examples
#' \dontrun{
#' boot_net(g, n = 500)
#'
#' }
#'
#' @export
#'


boot_network <- function(g, n = 500, left_ci = 0.005, right_ci = 0.995){

  if (inherits(g, "skynet")){
    g <- g[[1]]
  }else{
    g <- g
  }

      avg <- c() # Creates empty vector for output
      trv <- c()
      btw <- c()
  for(i in 1:n){
    gBoot <- g %>%
      #rewires graph g 10 times the amount of edges in g
      rewire(keeping_degseq(loops = FALSE, niter = ecount(g)*10))

    # Runs and saves statistics of interest
    avg[i] <- average.path.length(gBoot)
    trv[i] <- transitivity(gBoot)
    btw[i] <- mean(betweenness(gBoot))
  }
      # Calculates mean values and CIs
      avg <- c(quantile(avg, left_ci), quantile(avg, right_ci),
                        mean_random = mean(avg),
                        mean_empirical = average.path.length(g))

      trv <- c(quantile(trv, left_ci), quantile(trv, right_ci),
                 mean_random = mean(trv),
                 mean_empirical = transitivity(g))

      btw <- c(quantile(btw, left_ci), quantile(btw, right_ci),
               mean_random = mean(btw),
               mean_empirical = mean(betweenness(g)))

      bootstats <- t(data.frame(average.path.length = avg,
                                transitivity = trv,
                                betweenness = btw))
      bootstats <- as.data.frame(bootstats, col.names = colnames(bootstats))

      return(bootstats)
}

bootnet <- function(...){
  warning(paste("bootnet is deprecated, use boot_network(), instead."))
  do.call(boot_network, list(...))
}