hexbin

hexbin(xs, ys; kwargs...)

Plots a heatmap with hexagonal bins for the observations xs and ys.

Attributes

Specific to Hexbin

  • bins = 20: If an Int, sets the number of bins in x and y direction. If a Tuple{Int, Int}, sets the number of bins for x and y separately.

  • cellsize = nothing: If a Real, makes equally-sided hexagons with width cellsize. If a Tuple{Real, Real} specifies hexagon width and height separately.

  • threshold::Int = 1: The minimal number of observations in the bin to be shown. If 0, all zero-count hexagons fitting into the data limits will be shown.

  • scale = identity: A function to scale the number of observations in a bin, eg. log10.

Generic

  • colormap::Union{Symbol, Vector{<:Colorant}} = :viridis

  • colorrange::Tuple(<:Real,<:Real} = Makie.automatic sets the values representing the start and end points of colormap.

Examples

Setting the number of bins

Setting bins to an integer sets the number of bins to this value for both x and y. The minimum number of bins in one dimension is 2.

using CairoMakie

using Random
Random.seed!(1234)

f = Figure(resolution = (800, 800))

x = rand(300)
y = rand(300)

for i in 2:5
    ax = Axis(f[fldmod1(i-1, 2)...], title = "bins = $i", aspect = DataAspect())
    hexbin!(ax, x, y, bins = i)
    wireframe!(ax, Rect2f(Point2f.(x, y)), color = :red)
    scatter!(ax, x, y, color = :red, markersize = 5)
end

f

You can also pass a tuple of integers to control x and y separately.

using CairoMakie

using Random
Random.seed!(1234)

f = Figure(resolution = (800, 800))

x = rand(300)
y = rand(300)

for i in 2:5
    ax = Axis(f[fldmod1(i-1, 2)...], title = "bins = (3, $i)", aspect = DataAspect())
    hexbin!(ax, x, y, bins = (3, i))
    wireframe!(ax, Rect2f(Point2f.(x, y)), color = :red)
    scatter!(ax, x, y, color = :red, markersize = 5)
end

f

Setting the size of cells

You can also control the cell size directly by setting the cellsize keyword. In this case, the bins setting is ignored.

The height of a hexagon is larger than its width. This is why setting the same size for x and y will result in uneven hexagons.

using CairoMakie

using Random
Random.seed!(1234)

f = Figure(resolution = (800, 800))

x = rand(300)
y = rand(300)

for (i, cellsize) in enumerate([0.1, 0.15, 0.2, 0.25])
    ax = Axis(f[fldmod1(i, 2)...], title = "cellsize = ($cellsize, $cellsize)", aspect = DataAspect())
    hexbin!(ax, x, y, cellsize = (cellsize, cellsize))
    wireframe!(ax, Rect2f(Point2f.(x, y)), color = :red)
    scatter!(ax, x, y, color = :red, markersize = 5)
end

f

To get evenly sized hexagons, set the cell size to a single number. This number defines the cell width, the height will be computed as 2 * step_x / sqrt(3). Note that the visual appearance of the hexagons will only be even if the x and y axis have the same scaling, which is why we use aspect = DataAspect() in these examples.

using CairoMakie

using Random
Random.seed!(1234)

f = Figure(resolution = (800, 800))

x = rand(300)
y = rand(300)

for (i, cellsize) in enumerate([0.1, 0.15, 0.2, 0.25])
    ax = Axis(f[fldmod1(i, 2)...], title = "cellsize = $cellsize", aspect = DataAspect())
    hexbin!(ax, x, y, cellsize = cellsize)
    wireframe!(ax, Rect2f(Point2f.(x, y)), color = :red)
    scatter!(ax, x, y, color = :red, markersize = 5)
end

f

Hiding hexagons with low counts

All hexagons with a count lower than threshold will be removed:

using CairoMakie

using Random
Random.seed!(1234)

f = Figure(resolution = (800, 800))

x = randn(100000)
y = randn(100000)

for (i, threshold) in enumerate([1, 10, 100, 500])
    ax = Axis(f[fldmod1(i, 2)...], title = "threshold = $threshold", aspect = DataAspect())
    hexbin!(ax, x, y, cellsize = 0.4, threshold = threshold)
end
f

Changing the scale of the number of observations in a bin

You can pass a scale function to via the scale keyword, which will be applied to the bin counts before plotting.

using CairoMakie

using Random
Random.seed!(1234)

x = randn(100000)
y = randn(100000)

f = Figure()
hexbin(f[1, 1], x, y, bins = 40,
    axis = (aspect = DataAspect(), title = "scale = identity"))
hexbin(f[1, 2], x, y, bins = 40, scale=log10,
    axis = (aspect = DataAspect(), title = "scale = log10"))
f

Showing zero count hexagons

By setting threshold = 0, all hexagons that fit into the limits of the input data are shown. In this example, we add a transparent color to the start of the colormap and stroke each hexagon so the empty hexagons are visible but not too distracting.

using CairoMakie

using DelimitedFiles


a = map(Point2f, eachrow(readdlm(assetpath("airportlocations.csv"))))

f, ax, hb = hexbin(a,
    cellsize = 6,
    axis = (; aspect = DataAspect()),
    threshold = 0,
    colormap = [Makie.to_color(:transparent); Makie.to_colormap(:viridis)],
    strokewidth = 0.5,
    strokecolor = :gray50,
    scale = Makie.pseudolog10)

tightlimits!(ax)

Colorbar(f[1, 2], hb,
    label = "Number of airports",
    ticks = (0:3, ["0", "10", "100", "1000"]),
    height = Relative(0.5)
)
f