R/parameter_bootstraps.R

In cboettig/earlywarning: Detection of Early Warning Signals for Catastrophic Bifurcations

#' extract a data.frame of parameters from the bootstrapped replicate comparisons 
#' @param reps a list of outputs from compare()
#' @return a data.frame of columns "value" (parameter value), 
#' "parameter" (name of the parameter), "fit", a double-letter code indicating
#' first which model created the simulation and second which model was fit to the data,
#' "rep", a replicate id number.  
#' @details Consider ggplot2 tools to visualize this data, or use the subset() function 
#' on this data set to look at a particular parameter from a particular model fit.
#' see examples for illustrations.  
#' @seealso \link{compare}, \link{roc_data}, \link{lik_ratios}, \link{subset}
#' @import reshape2
#' @export
#' @examples
#' data("example_analysis")
#' pars <- parameter_bootstraps(reps)
#' subset(pars, fit=="BB" && parameter=="m")
#' require(ggplot2)
#' require(Hmisc)
#' ggplot(subset(pars, fit %in% c("AA", "BB")), aes(parameter, value, fill = fit)) +
#'   stat_summary(fun.y = mean, geom = "bar", position = "dodge") +
#'   stat_summary(fun.data = median_hilow, geom = "pointrange", aes(color = fit), 
#'               position = position_dodge(width = 0.90), conf.int = 0.95)

parameter_bootstraps <- function(reps){
  dat <- lapply(reps, function(rep){
          list(AA=as.list(fitted(rep$AA)), AB=as.list(fitted(rep$AB)), 
               BA=as.list(fitted(rep$BA)), BB=as.list(fitted(rep$BB)))
  })
  dat <- melt(dat)
  names(dat) <- c("value", "parameter", "fit", "rep") 
  dat
}