Gadfly is a plotting and data visualization system written in Julia.

It's influenced heavily by Leland Wilkinson's book The Grammar of Graphics and Hadley Wickham's refinment of that grammar in ggplot2.

It renders publication quality graphics to PNG, Postscript, PDF, SVG, and Javascript. The Javascript backend uses d3 to add interactivity like panning, zooming, and toggling.

Check out the manual for more details and examples.

From the Julia REPL a reasonably up to date version can be installed with

Pkg.add("Gadfly")

This will likely result in half a dozen or so other packages also being installed.

Gadfly works best with the C libraries cairo, pango, and fontconfig installed. The PNG, PS, and PDF backends require cairo, but without it the SVG and Javascript/D3 backends are still available.

Complex layouts involving text are also somewhat more accurate when pango and fontconfig are available.

Julia's Cairo bindings can be installed with

Pkg.add("Cairo")

All interaction with Gadfly is through the plot function, which takes three major forms.

plot(data::AbstractDataFrame, elements::Element...; mapping...)

This form is the standard "grammar of graphics" method of plotting. data is supplied in the form of a dataframe, columns of the data are bound to aesthetics, and plot elements including scales, coordinates, statistics, guides, and geometries are added to the plot.

All of the examples that follow will be plotting data from RDatasets.

To render these plots to a file, call draw on the resulting plot.

draw(SVG("myplot.svg", 6inch, 3inch), plot(...))

A few examples now.

# E.g.
plot(data("datasets", "iris"),x="Sepal.Length", y="Sepal.Width", Geom.point)

# E.g.
plot(data("car", "SLID"), x="wages", color="language", Geom.bar)

A catalog of plot elements given later in this document.

plot(elements::Element...; mapping...)

Along with the orthodox invocation of plot, some relaxed invocations of the grammar exist as a "slang of graphics". This form of plot omits the the data frame. Instead, plain old arrays are bound to aesthetics.

# E.g.
plot(x=collect(1:100), y=sort(rand(100)))

If no geometry is specified, like in the example above, a Geom.point is stuck into your plot.

This plot otherwise works the same. We might want to name these axis, for example.

# E.g.
plot(x=collect(1:100), y=sort(rand(100)),
     Guide.XLabel("Index"), Guide.YLabel("Step"))

plot(f::Function, a, b, elements::Element...)

plot(fs::Array, a, b, elements::Element...)

@plot(expr, a, b)

Some special forms of plot exist for quickly generating 2d plots of functions.

# E.g.
plot([sin, cos], 0, 25)

# E.g.
@plot(cos(x)/x, 5, 25)

Plot elements in Gadfly are statistics, scales, geometries, and guides. Each operates on data bound to aesthetics, but in different ways.

Statistics are functions taking as input one or more aesthetics, operating on those values, then output to one or more aesthetic. For example, drawing of boxplots typically uses the boxplot statistic (Stat.boxplot) that takes as input the x and y aesthetic, and outputs the middle, and upper and lower hinge, and upper and lower fence aesthetics.

Scales, similarly to statistics apply a transformation to the original data, typically mapping one aesthetic to the same aesthetic, while retaining the original value. The Scale.x_log10 aesthetic maps the x aesthetic back the x aesthetic after applying a log10 transformation, but keeps track of the original value so that data points are properly identified.

Finally geometries are responsible for actually doing the drawing. A geometry takes as input one or aesthetics, and used data bound to these aesthetics to draw things. The Geom.point geometry draws points using the x and y aesthetics, the Geom.lines geometry draws lines, and so on.

Very similar to geometries are guides, which draw graphics supported the actual visualization, such al axis ticks and labels and color keys. The major distinction is that geometries always draw within the rectangular plot frame, while guides have some special layout considerations.

Gadfly plots can be rendered to number of formats. Without cairo, or any non-julia libraries, it can produce SVG and d3-powered javascript. Installing cairo gives you access to the PNG, PDF, and PS backends. Rendering to a backend works the same for any of these.

some_plot = plot(x=[1,2,3], y[4,5,6])
draw(PNG("myplot.png", 6inch, 3inch), some_plot)

The D3 backend writes javascript. Making use of it's output is slightly more involved than with the image backends.

Rendering to Javascript is easy enough:

draw(D3("mammals.js", 6inch, 6inch), p)

Before the output can be included, you must include the d3 and gadfly javascript libraries. The necessary include for Gadfly is "gadfly.js" which lives in the src directory (which you can find by running joinpath(Pkg.dir("Gadfly"), "src", "gadfly.js") in julia).

D3 can be downloaded from here.

Now the output can be included in an HTML like.

<script src="d3.min.js"></script>
<script src="gadfly.js"></script>

<!-- Placed whereever you want the graphic to be rendered. -->
<div id="my_chart"></div>
<script src="mammals.js"></script>
<script>
draw("#my_chart");
</script>

A div element must be placed, and the draw function defined in mammals.js must be passed the id of this element, so it knows where in the document to place the plot.

This is a new and fairly complex piece of software. Filing an issue to report a bug, counterintuitive behavior, or even to request a feature is extremely valuable in helping me prioritize what to work on, so don't hestitate.