dogs.R

# Reading in 'dogs.csv'
# Created by Alex Bresler, available at:
# https://github.com/abresler/abresler.github.io/blob/master/blog/2015/february/exploring_agility_show/data/dogs.csv
dogs <- read.csv("dogs.csv", header = TRUE, stringsAsFactors = FALSE)

# Setting a numeric id
node_id <- 1:nrow(dogs)

# Getting unique values for the 'breed' column
unique_node_2 <- unique(dogs$breed)

# Getting secondary node id values from arbitrary column (breed in this case)
node_id_2 <- vector(mode = "numeric")
for (i in 1:nrow(dogs)){
  node_id_2 <- c(node_id_2, which(unique_node_2 %in% dogs$breed[i]))
}

top_five_id_2 <- sort(table(node_id_2), decreasing = TRUE)[1:5]

# Setting the maximum height of a scaled node
maxheight <- 0.5

# Scaling the maximum height of the node
height <- dogs$height_class/max(dogs$height_class) * maxheight

# Adding new columns to the 'dogs' data frame, because 'height' is included
# it'll be applied as a parameter to each node staement
dogs <- cbind(node_id, height, dogs, node_id_2)

# Creating node statements for the primary nodes (all dogs)
node_statements_1 <- graphviz_node_edge_blocks_df(nodes = dogs)

# Creating node statements for the secondary nodes (all breeds of dogs)
node_statements_2 <-
  paste(paste0("  node [peripheries = 2",
               ifelse(unique(node_id_2) %in% as.numeric(names(top_five_id_2)),
                      paste0(", label = '", unique_node_2[unique(node_id_2)], "']"), "]"),
               "'2_", unique(node_id_2), "' \n"), collapse = "; ")

# Creating edge statements (from each dog to the breed of dog)
edge_statements_1 <- paste(gsub("\\+", "",
                                paste0("'", "1_", dogs$node_id, "' -> '",
                                       "2_", dogs$node_id_2, "'")),
                           collapse = "\n")

# Let's make a 'twopi' styled graph
grViz("
digraph dogs {

  # Graph statements
  graph [layout = twopi, overlap = false]

  # Node statements
  node [shape = circle, label = '', style = filled, color = lightblue]
  @@1
  node [height = 1, label = '', style = filled, color = turquoise]
  @@2

  # Edge statements
  edge [color = gray, arrowsize = 0.5]
  @@3

}
[1]: node_statements_1
[2]: node_statements_2
[3]: edge_statements_1

")