R/plot_workers.R

In danielhorn/multicrit_result_test: Multi-objective selection of algorithm portfolios

# Ground Zero Plot, Chaos Plot
# Plot The Raw Data, Color for algo, x and y for the variables and facet wrap for repl
plotChaosPlot = function(data, var.cols, algo.col, repl.col, data.col, colors) {
  # Split dataset into it replications
  data.splitted = split(data, data[, c(repl.col, algo.col, data.col)])
  
  # Apply Pareto-Filt
  data.splitted = lapply(data.splitted, function(d) {
    if (nrow(d) == 0L)
      d
    else
      d[nds_rank(as.matrix(t(d[, var.cols]))) == 1L, ]
  })
  
  # remerge data
  data = do.call(rbind, data.splitted)
  
  p = ggplot(data, ggplot2::aes_string(var.cols[1], var.cols[2], colour = algo.col)) +
    geom_point(size = 3) + ggtitle("All Pareto Optimal Points") +
    scale_colour_manual(values = colors) + facet_grid(reformulate(repl.col, data.col))
  return(p)
}

# First, Third and Fifth Plot: EAF plot.
plotEAF = function(data, var.cols, algo.col, repl.col, data.col, colors) {
  fronts = data[, c(var.cols, algo.col, repl.col, data.col)]
  fronts[, algo.col] = factor(fronts[, algo.col])
  fronts[, repl.col] = factor(fronts[, repl.col])
  
  p = eafGGPlot(data = fronts, var.names = var.cols, group.name = algo.col,
    replication.name = repl.col, data.name = data.col, percentiles = 50) +
    geom_line(size = 2) +  scale_colour_manual(values = c(colors)) +
    guides(linetype = FALSE)
  return(p)
}

# Second Plot: Plot in how many repls an algo has non.dom points, via barplot
plotDominationSelection = function(data, colors, repl.count, algo.col, eta) {
  
  data.long = data.table::melt(data)
  names(data.long) = c("algo", "dataset", "count")
  
 # data.long = unlist(lapply(1:length(data), function(i) rep(names(data)[i], data[i])))
#  data.long = data.frame(algo = factor(data.long, levels = names(data)))
   
  p = ggplot(data.long, aes_string(x = "algo", y = "count", fill = "algo")) +
    geom_bar(stat="identity") +
    scale_fill_manual(values = colors, drop = FALSE, name = algo.col) +
    scale_x_discrete(drop = FALSE) +
    geom_hline(yintercept = repl.count * eta, size = 1, alpha = 0.5) +
    facet_wrap("dataset")
  return(p)
}