# Plotting

Most interaction with Gadfly is through the plot function. Plots are described by binding data to aesthetics, and specifying a number of elements including Scales, Coordinates, Guides, and Geometries. Aesthetics are a set of special named variables that are mapped to a geometry. How this mapping occurs is defined by the elements.

This "grammar of graphics" approach tries to avoid arcane incantations and special cases, instead approaching the problem as if one were drawing a wiring diagram: data is connected to aesthetics, which act as input leads, and elements, each self-contained with well-defined inputs and outputs, are connected and combined to produce the desired result.

## Functions and Expressions

Along with the standard plot methods operating on DataFrames and Arrays described in the Tutorial, Gadfly has some special signatures to make plotting functions and expressions more convenient.

plot(f::Function, lower, upper, elements...; mapping...)
plot(fs::Vector{T}, lower, upper, elements...; mapping...) where T <: Base.Callable
plot(f::Function, xmin, xmax, ymin, ymax, elements...; mapping...)
spy(M::AbstractMatrix, elements...; mapping...) -> Plot

For example:

p1 = plot([sin,cos], 0, 2pi)
p2 = plot((x,y)->sin(x)+cos(y), 0, 2pi, 0, 2pi)
p3 = spy(ones(33)*sin.(0:(pi/16):2pi)' + cos.(0:(pi/16):2pi)*ones(33)')
hstack(p1,p2,p3)

## Adding to a plot

Another feature is that a plot can be added to incrementally, using push!.

p = plot()
push!(p, layer(x=[2,4], y=[2,4], size=[1.4142], color=[colorant"gold"]))
push!(p, Coord.cartesian(fixed=true))
push!(p, Guide.title("My Awesome Plot"))

## Wide-formatted data

Gadfly is designed to plot data in so-called "long form", in which data that is of the same type, or measuring the same quantity, are stored in a single column, and any factors or groups are specified by additional columns. This is how data is typically stored in a database.

Sometimes data tables are organized by grouping values of the same type into multiple columns, with a column name used to distinguish the grouping. We refer to this as "wide form" data.

To illustrate the difference consider some historical London birth rate data.

births = RDatasets.dataset("HistData", "Arbuthnot")[:,[:Year, :Males, :Females]]
RowYearMalesFemales
1162952184683
2163048584457
3163144224102
4163249944590
5163351584839
6163450354820
............

This table is wide form because "Males" and "Females" are two columns both measuring number of births. Wide form data can always be transformed to long form (e.g. with the stack function in DataFrames) but this can be inconvenient, especially if the data is not already in a DataFrame.

stack(births, [:Males, :Females])
RowvariablevalueYear
1Males52181629
2Males48581630
3Males44221631
............
162Females76231708
163Females73801709
164Females72881710

The resulting table is long form with number of births in one column, here with the default name given by stack: "value". Data in this form can be plotted very conveniently with Gadfly.

plot(stack(births, [:Males, :Females]), x=:Year, y=:value, color=:variable,
Geom.line)