Is there a way to make a stacked histogram in Makie like Stacked histogram on a log scale — seaborn 0.13.2 documentation?
The documentation only has examples about multiple histograms.
Maybe @jling knows how one can do that in FHist.jl?
If there is no way to do it in Makie directly I can write a patch for Makie.
Something along these lines
using CairoMakie, StatsBase
function stacked_hist!(ax, datasets; bins, colors)
@assert length(datasets) == length(colors)
edges = collect(bins)
counts = [fit(Histogram, d, edges).weights for d in datasets]
M = hcat(counts...)
C = cumsum(M; dims=2)
width = edges[2] - edges[1]
centers = @view(edges[1:end-1]) .+ width/2
for i in eachindex(centers)
bottom = 0.0
for j in 1:size(C,2)
top = C[i,j]
# rectangle polygon corners
x0 = centers[i] - width/2
x1 = centers[i] + width/2
poly!(ax,
Point2f[(x0,bottom), (x1,bottom), (x1,top), (x0,top)],
color = colors[j])
bottom = top
end
end
ax
end
f = Figure()
ax = Axis(f[1,1], xlabel="Value", ylabel="Count")
stacked_hist!(ax,
[randn(1000), randn(1000) .+ 1, randn(1000) .+ 2];
bins = -4:0.5:6,
colors = [:tomato, :dodgerblue, :seagreen]
)
f
1 Like
I wrote a PR for multi-histogram plot in Makie:
1 Like
I’m pretty sure AoG can do this, see the third plot in the histogram docs: Analyses | AlgebraOfGraphics
specs = data(df) * mapping(:x, stack=:z, color=:z) * histogram()
draw(specs)
2 Likes