R/functions_visualisation.R

In ERPM: Exponential Random Partition Models

######################################################################
## Simulation and estimation of Exponential Random Partition Models ##
## Visualization of paritions                                       ##
## Author: Alexandra Amani                                          ##
######################################################################


#' Visualization of partition
#'
#'This function plot the groups of a partition
#'
#'
#' @importFrom igraph graph_from_adjacency_matrix
#' @importFrom igraph make_clusters
#' @importFrom igraph layout.auto
#' @importFrom RColorBrewer brewer.pal
#' @param partition A partition (vector)
#' @param title Character, the title of the plot (default=NULL)
#' @param group.color A vector with the colors of the groups (default=NULL)
#' @param attribute.color A vector, attribute to represent with colors (default=NULL)
#' @param attribute.shape A vector, attribute to represent with shapes (default=NULL)
#' @return A plot of the partition
#' @importFrom grDevices rgb
#' @examples
#' p <- c(1,1,1,2,2,2,2,3,3,3,4,4,4,4,4,4)
#' attr1 <- c(1,0,0,1,0,0,1,0,1,0,1,1,1,1,1,2)
#' attr2 <- c(1,1,1,1,0,0,3,0,1,0,1,1,1,1,1,2)
#' plot_partition(p,attribute.color = attr1, attribute.shape = attr2)
#' @export

plot_partition <- function(partition, title = NULL, group.color = NULL,
                           attribute.color = NULL, attribute.shape = NULL){

  num.nodes <- length(partition)
  num.group <- table(partition)
  cat.col <- unique(attribute.color)
  possible.color <- brewer.pal(length(cat.col), name = "Reds")
  possible.shape <-  c( "circle","square","csquare","rectangle","crectangle" ,"vrectangle")

     # Color for Attribute
    if (is.null(attribute.color)){
          n.color <- 'black'
    }else{
          n.color <- attribute.color
          for (i in 1: length(cat.col)){
          n.color <- replace(n.color,n.color==cat.col[i],possible.color[i])
      }
    }


    # Shape for Attribute
    if (is.null(attribute.shape)){
          n.shape <- 'circle'
    }else{
      n.shape <- attribute.shape
          cat.shape <- unique(attribute.shape)
          for (i in 1:length(cat.shape)){
          n.shape <- replace(n.shape,n.shape==cat.shape[i],possible.shape[i])
      }
    }


    # set colors for groups
        if (is.null(group.color)) {
          g.colors <- brewer.pal(max(partition), name = "Blues")
        }else {
          g.colors = group.color
        }

    # make graph and clusters for igraph
    affiliation <- as.matrix(table(data.frame(actor = 1:num.nodes, group= partition)))
    adjacency <- affiliation %*% t(affiliation)
    diag(adjacency) <- 0
    graph <- graph_from_adjacency_matrix(adjacency)
    clusters <- make_clusters(graph, membership = partition, modularity = FALSE)
    layout <- layout.auto(graph)

    # Plot

    plot(clusters, graph,
         layout=layout,
         col = n.color,
         vertex.shape = n.shape,
         vertex.label=NA,
         vertex.size=5,
         edge.color = rgb(1,0,0,alpha = 0), # transparent edges
         mark.border = "white", # color for group border
         mark.col = g.colors,
         main = title) # color for group
}