R/fviz-mca-var-corr.R

In inventionate/TimeSpaceAnalysis: Statistical tools for time-space analysis

#' @include utilities.R
NULL
#' Visualize MCA variable representation square.
#'
#' @param res_gda MCA result.
#' @param axes axes to plot.
#' @param geom whether points or labels to plot.
#' @param labelsize size of labels.
#' @param pointsize size of points.
#' @param invisible hide "passive" or "active" variables.
#' @param labels label points or not (boolean).
#' @param repel repel labels (boolean).
#' @param select selection of variables (names) or eta2 values (all above value).
#' @param title plot title.
#' @param plot_modif_rates plot modified rates instead of eigenvalue percentage (boolean).
#'
#' @return ggplot2 visualization of variable correlation square (variables representation).
#' @export
fviz_mca_var_corr <- function(res_gda, axes = c(1,2), geom=c("point", "text"), labelsize = 4, pointsize = 2,
                              invisible = NULL, labels = TRUE, repel = TRUE, select = list(name = NULL, eta2 = NULL),
                              plot_modif_rates = TRUE, title = "MCA - Variable Representation") {

  vars <- get_mca_var_corr(res_gda, axes = axes)

  colnames(vars)[3:4] <-  c("x", "y")

  # Selection
  if (!is.null(select)) vars <- .select(vars, select)

  # Exclude invisible Data
  vars <-
    vars[which(Hmisc::`%nin%`(vars$type, invisible)), , drop = TRUE] %>%
    data.frame()

  # Plot
  p <- ggplot(data = vars)

  if ("point" %in% geom) {
    p <-
      p +
      geom_point(
        aes(x, y, colour = type, shape = type),
        size = pointsize)
  }

  if (labels & "text" %in% geom) {
    if(repel) {
      p <-
        p +
        ggrepel::geom_text_repel(
          mapping = aes(x, y, color = type, label = name),
          size = labelsize
        )
    } else {
      p <-
        p +
        geom_text(
          mapping = aes(x, y, color = type, label = name),
          size = labelsize, nudge_y = -0.015
        )
    }
  }

  # Set fix dimensions
  p <-
    add_theme(p) +
    scale_x_continuous(expand = c(0,0), limits = c(0,1), labels = comma) +
    scale_y_continuous(expand = c(0,0), limits = c(0,1), labels = comma)

  # If there are no passive variables use only one colour
  if (nlevels(vars$type) == 1) {
    p <- p + scale_colour_manual(values = c("black"))
  }

  # Update labels
  p <- .gda_plot_labels(res_gda, p, title, axes, plot_modif_rates)

  # Return ggplot
  p
}