riverplot: Create a Sankey plot

In ischeinfeld/riverplot: Sankey or Ribbon Plots

Description

Create a Sankey plot

Usage

## S3 method for class 'riverplot'
plot(x, ...)

riverplot(x, direction = "lr", lty = 0, default_style = NULL,
  gravity = "top", node_margin = 0.1, nodewidth = 1.5, plot_area = 0.5,
  nsteps = 50, add_mid_points = NULL, xscale = 1, yscale = "auto",
  mar = c(0, 0, 0, 0), add = FALSE, usr = NULL, fix.pdf = FALSE, ...)

Arguments

x	An object of class riverplot
...	any further parameters passed to riverplot() are appended to the default style
direction	"lr" (left to right) or "rl" (right to left)
lty	Line style to use
default_style	default graphical style
gravity	how the nodes are placed vertically. No effect if node vertical positions are specified via node_ypos member
node_margin	how much vertical space should be kept between the nodes
nodewidth	width of the node (relative to font size)
plot_area	fraction of vertical space to be used as main plot area
nsteps	number of interpolating steps in drawing the segments
add_mid_points	attempt to get a smoother plot by adding additional nodes. Set this parameter to FALSE if you are setting node vertical position manually. If add_mid_points is equal to NULL (the default), then the mid points are added only if node_ypos is empty.
xscale	scale the positions of the nodes by that factor. This can be used to "squeeze" the diagram to the left as necessary.
yscale	scale the edge width values by multiplying with this factor. If yscale is equal to "auto", scaling is done automatically such that the vertical size of the largest node is approximately equal to 15 If no node_ypos is specified in the riverplot object, no scaling is done. If yscale is equal to 1, no scaling is done.
mar	margins to set (as accepted by par(mar=..)). Set to NULL if you want the margins untouched.
add	If TRUE, do not call plot.new(), but add to the existing plot.
usr	coordinates at which to draw the plot in form (x0, x1, y0, y1).
fix.pdf	Try to fix PDF output if it looks broken (with thin white lines). Don't use this option if you are using transparent colors.

Details

This functions create a Sankey plot given a riverplot object (plot is just a wrapper for the riverplot function. The object to be drawn is a list specifying the plot; see the makeRiver function for exact specifications and the riverplot.example to see how it can be created. Whether or not the list used to plot is exactly of class riverplot-class does not matter as long as it has the correct contents.

Style information which is missing from the riverplot object x (for example, if the node style is not specified for each node in the object) is taken from the default.style parameter. See functions default.style() and updateRiverplotStyle() to learn how to create and modify the styles.

Whether or not the list used to plot is exactly of class riverplot-class does not matter as long as it has the correct contents. These functions here are for the convenience of checking that

The nodes are drawn from bottom to top in the order they are found in the riverplot object. There is no clever algorithm for placing the nodes minimizing the number of crossing edges yet; you need to manipulate the object directly to achieve the desired effect.

Value

riverplot return invisibly a matrix containing the actual positions (in user coordinates) of the nodes drawn on the screen. Note that it also may contain additional, invisible nodes that have been created by the algorithm to better fit on the screen.

Known problems

There is a problem with transparency and PDFs. In short, if you try to save your riverplot graphics as PDF, you will observe thin, white vertical lines everywhere on the curves. The reasons for that are unclear, but have something to do with PDF rendering (if you generate EPS, the output looks good).

There is a kind of fix to that: use the fix.pdf=TRUE option. Unfortunately, this solution does not work if you use transparent colors (you will have a different kind of vertical lines). Unfortunately, I don't have a solution for that problem yet.

See Also

default.style updateRiverplotStyle minard

Examples

x <- riverplot.example()
plot(x)
plot(x, srt=90, lty=1)

ischeinfeld/riverplot documentation built on May 13, 2019, 4:05 a.m.