Join 10350+ others. No spamming.
I promise!

Follow us at github.



rich-iannone / DiagrammeR


Graph and network visualization using tabular data in R


Travis-CI Build Status Issue Stats Issue Stats

With DiagrammeR, you can easily create graph diagrams. You can either use Markdown-like text to describe and render a diagram, or, use a collection of functions to create graph objects. The output can be viewed in the RStudio Viewer, it can be incorporated in R Markdown, and it can be integrated in shiny web apps. Because we are doing this in R we can and always should add much more R code into the mix.

Go to the project website and view a video walkthrough for a graph diagram that's created with a few lines of text and is just as easily customizable. After being all fired up from that intense video-tutorial extravaganza, have a look at the DiagrammeR Docs to learn more.

Creating Graphviz Graphs

It's possible to make graph diagrams using the Graphviz support included in package. Simply specify a valid Graphviz graph in the DOT language either in the form of a string, a reference to a Graphviz file (with a .gv file extension), or as a text connection.

Here is an example where nodes (in this case styled as rectangles and circles) can be easily defined along with their connections:


digraph {

  # graph attributes
  graph [overlap = true]

  # node attributes
  node [shape = box,
        fontname = Helvetica,
        color = blue]

  # edge attributes
  edge [color = gray]

  # node statements
  A; B; C; D; E
  F [color = black]

  # node attributes
  node [shape = circle,
        fixedsize = true,
        width = 0.9]

  # node statements
  1; 2; 3; 4; 5; 6; 7; 8

  # edge statements
  A->1; B->2                   // gray
  B->3 [color = red]           // red
  B->4                         // gray
  C->A [color = green]         // green
  1->D; E->A; 2->4; 1->5; 1->F // gray
  E->6; 4->6; 5->7; 6->7       // gray
  3->8 [color = blue]          // blue

There is a great variety of ways to style the nodes and edges in a Graphviz graph diagram. Also, the layout of a graph can changed by using the neato, twopi, and circo rendering engines.

Using DiagrammeR Functions to Define Graphs

In the last example you saw what was essentially a text string being passed into a single function. That's not very R-like, is it? Well, it's a good thing that there's a collection of graph functions available for creating and manipulating graphs (specifically, graph objects). They allow you to generate node and edge data frames (collections of nodes or edges along with their attributes), perform scaling of attribute values with data values, create graph objects, render those graphs, modify those graphs, get information from the graphs, create a series of graphs, and... so much more. A lot of the graph functions work together in a pipeline using magrittr pipes. Here is an example of an approach to building the simple graph above using DiagrammeR functions:


graph <-
  create_graph() %>%
  set_graph_name("Boxes and Circles") %>%
  set_graph_time() %>%
  set_global_graph_attr("graph", "overlap", "true") %>%
  set_global_graph_attr("node", "shape", "box") %>%
  set_global_graph_attr("node", "fontname", "Helvetica") %>%
  set_global_graph_attr("node", "color", "blue") %>%
  set_global_graph_attr("edge", "color", "gray") %>%
  add_node_df(create_nodes(c("A", "B", "C", "D", "E", "F"))) %>%
  set_node_attr("F", "color", "black") %>%
  add_node_df(create_nodes(1:8)) %>%
  select_nodes(1:8) %>%
  set_node_attr_with_selection("shape", "circle") %>%
  set_node_attr_with_selection("fixedsize", "true") %>%
  set_node_attr_with_selection("width", 0.9) %>%
  clear_selection() %>%
  add_edge_df(create_edges(c("A", "B", "B", "B",
                             "C", "1", "E", "2",
                             "1", "1", "E", "4",
                             "5", "6", "3"),
                           c("1", "2", "3", "4",
                             "A", "D", "A", "4",
                             "5", "F", "6", "6",
                             "7", "7", "8"))) %>%
  set_edge_attr("B", "3", "color", "red") %>%
  set_edge_attr("C", "A", "color", "green") %>%
  set_edge_attr("3", "8", "color", "blue")


With the graph-building functions, it's possible to generate a graph with data available in a data frame. The general idea is to build specialized data frames that contain either node data and attributes (node data frames) and those data frames that contain edge data and edge attributes (edge data frames). These data frames are permitted to have node and edge attributes and also columns of other data. Then we can incorporate the data into a graph object which can itself be subject to inspection and transformations. Because the attributes are always kept alongside the node and edge definitions (within the graph object itself), we can easily work with them and modify the values of the styling attributes and differentiate nodes and edges by size, color, shape, opacity, length, etc. Here are some of the available graph functions:

Here is a simple workflow for building and rendering a graph object:

Want to learn more? Head over to the DiagrammeR Docs to see plenty of examples and explanations.

An Example with Data from the nycflights13 Package

Using the flights dataset from the nycflights13 R package, we can create a graph diagram. Here, the green lines show flights that weren't late arriving at their destinations (red indicates those late arrivals).

The R code for generating this:


# Choose a day from 2013 for NYC flight data
# (You can choose any Julian day, it's interesting to see results for different days)
day_of_year <- 10 

# Get a data frame of complete cases (e.g., flights have departure and arrival times)
nycflights13 <-
  nycflights13::flights[which(complete.cases(nycflights13::flights) == TRUE), ]

# Generate a POSIXct vector of dates using the 'ISOdatetime' function
# Columns 1, 2, and 3 are year, month, and day columns
# Column 4 is a 4-digit combination of hours (00-23) and minutes (00-59)
date_time <-
  data.frame("date_time" =
               ISOdatetime(year = nycflights13[,1],
                           month = nycflights13[,2],
                           day = nycflights13[,3],
                           hour = gsub("[0-9][0-9]$", "", nycflights13[,4]),
                           min = gsub(".*([0-9][0-9])$", "\\1", nycflights13[,4]),
                           sec = 0, tz = "GMT"))

# Add the POSIXct vector 'date_time' to the 'nycflights13' data frame
nycflights13 <- cbind(date_time, nycflights13)

# Select flights only from the specified day of the year 2013
nycflights13_day <-
         date_time >= ymd('2013-01-01', tz = "GMT") + days(day_of_year - 1) &
           date_time < ymd('2013-01-01', tz = "GMT") + days(day_of_year))

# Create the node data frame
# Column 12 is the 3-letter code for the airport departing from
# Column 13 is for the airport arriving to
# (Option: change df to 'nycflights13_day' and only airports used for the day will be included)
nodes_df <- create_nodes(nodes = unique(c(nycflights13[,12],
                         label = FALSE)

# The edge data frame must have columns named 'from' and 'to'
# The color attribute is determined with an 'ifelse' statement, where
# column 8 is the minutes early (negative values) or minutes late (positive values)
# for the flight arrival
edges_df <- create_edges(from = nycflights13_day[,12],
                         to = nycflights13_day[,13],
                         color = ifelse(nycflights13_day[,8] < 0,
                                    "green", "red"))

# Set the graph diagram's default attributes for...

# ...nodes
node_attrs <- c("style = filled", "fillcolor = lightblue",
                "color = gray", "shape = circle", "fontname = Helvetica",
                "width = 1")

# ...edges
edge_attrs <- c("arrowhead = dot")

# ...and the graph itself
graph_attrs <- c("layout = circo",
                 "overlap = false",
                 "fixedsize = true",
                 "ranksep = 3",
                 "outputorder = edgesfirst")

# Generate the graph diagram in the RStudio Viewer.
create_graph(nodes_df = nodes_df, edges_df = edges_df,
               graph_attrs = graph_attrs, node_attrs = node_attrs,
               edge_attrs = edge_attrs, directed = TRUE) %>%
  render_graph(width = 1200, height = 800)


DiagrammeR is used in an R environment. If you don't have an R installation, it can be obtained from the Comprehensive R Archive Network (CRAN). It is recommended that RStudio be used as the R IDE to take advantage of its rendering capabilities and the code-coloring support for Graphviz and mermaid diagrams.

You can install the development version of DiagrammeR from GitHub using the devtools package.


Or, get the v0.8.1 release from CRAN.