fc_plot.CS.R

# fc_plot.CS.R
# Plotting case stability from list generated by CS.adj()

plot.CS <- function(stability_list, 
                    tit = "Centrality stability") {
  df_list <- list()
  
  for (i in seq_along(stability_list)) {
    # Capturing proportions from element name
    raw_name <- names(stability_list)[i]
    proportion_value <- as.numeric(gsub("prop_", "", raw_name))
    
    # Building data.frame
    metrics_df <- stability_list[[i]]
    
    # Iterating by metric (column)
    for (metric_name in colnames(metrics_df)) {
      values <- metrics_df[[metric_name]]
      mean_val <- mean(values, na.rm = TRUE)
      se <- sd(values, na.rm = TRUE) / sqrt(sum(!is.na(values)))
      ci <- 1.96 * se
      
      df_list[[length(df_list) + 1]] <- data.frame(
        Proportion = proportion_value,
        Metric = metric_name,
        MeanCorrelation = mean_val,
        CI_Lower = mean_val - ci,
        CI_Upper = mean_val + ci
      )
    }
  }
  
  # Merging data.frames in a sigle one
  df <- do.call(rbind, df_list)
  
  # Plot
  ggplot(df, aes(x = Proportion * 100)) +
    geom_ribbon(aes(ymin = CI_Lower, ymax = CI_Upper, fill = Metric), alpha = 0.2) +
    geom_line(aes(y = MeanCorrelation, color = Metric), linewidth = 1.2) +
    geom_point(aes(y = MeanCorrelation, color = Metric), size = 2) +
    geom_hline(yintercept = 0.7, linetype = "dashed", color = "gray40", linewidth = 1) +
    scale_x_continuous(trans = "reverse", breaks = sort(unique(df$Proportion * 100), decreasing = TRUE)) +
    coord_cartesian(ylim = c(0, 1)) +
    labs(
      title = tit,
      x = "Sampled cases (%)",
      y = "Average correlation with original sample"
    ) +
    theme_minimal() +
    theme(legend.position = "top")
}