An alluvial/Sankey-style diagram showing how opinions or group memberships shift between survey waves or time periods. Flows connect the same response categories across columns, revealing patterns of opinion change, stability, and polarization. Unlike a basic alluvial diagram, this variant emphasizes distinguishing stable respondents from net changers and displays respondent totals per category at each wave.
# anyplot.ai
# alluvial-opinion-flow: Opinion Flow Diagram
# Library: makie 0.21.9 | Julia 1.11.9
# Quality: 90/100 | Created: 2026-08-24
using CairoMakie
using Colors
using Random
# Theme tokens (see prompts/default-style-guide.md "Background" + "Theme-adaptive Chrome")
THEME = get(ENV, "ANYPLOT_THEME", "light")
PAGE_BG = THEME == "light" ? colorant"#FAF8F1" : colorant"#1A1A17"
INK = THEME == "light" ? colorant"#1A1A17" : colorant"#F0EFE8"
INK_SOFT = THEME == "light" ? colorant"#4A4A44" : colorant"#B8B7B0"
# Imprint categorical palette — canonical order, position 1 always brand green
IMPRINT_PALETTE = [
colorant"#009E73", # 1 — brand green
colorant"#C475FD", # 2 — lavender
colorant"#4467A3", # 3 — blue
colorant"#BD8233", # 4 — ochre
colorant"#AE3030", # 5 — matte red
]
# Data — quarterly political opinion survey, 1000 respondents tracked across 4 waves
Random.seed!(42)
waves = ["Q1 2025", "Q2 2025", "Q3 2025", "Q4 2025"]
categories = ["Strongly Agree", "Agree", "Neutral", "Disagree", "Strongly Disagree"]
n_wave = length(waves)
n_cat = length(categories)
wave_counts = Vector{Vector{Int}}(undef, n_wave)
wave_counts[1] = [230, 260, 220, 180, 110]
# Diagonal-heavy transition matrix: most respondents stay put, some drift to
# neighboring categories, few make large jumps — a realistic opinion-survey pattern.
base_transition = [
0.78 0.14 0.05 0.02 0.01
0.10 0.70 0.14 0.05 0.01
0.03 0.12 0.68 0.12 0.05
0.01 0.05 0.14 0.68 0.12
0.01 0.02 0.05 0.14 0.78
]
flow_counts = Vector{Matrix{Int}}(undef, n_wave - 1)
for w in 1:(n_wave - 1)
noise = 0.03 .* (rand(n_cat, n_cat) .- 0.5)
transition = max.(base_transition .+ noise, 0.0)
transition = transition ./ sum(transition; dims = 2)
flows = zeros(Int, n_cat, n_cat)
for s in 1:n_cat
remaining = wave_counts[w][s]
for t in 1:(n_cat - 1)
take = min(round(Int, wave_counts[w][s] * transition[s, t]), remaining)
flows[s, t] = take
remaining -= take
end
flows[s, n_cat] = remaining
end
flow_counts[w] = flows
next_counts = [sum(flows[:, t]) for t in 1:n_cat]
wave_counts[w + 1] = next_counts
end
# Layout — stack category nodes per wave column; node_gap reserves whitespace
# above each node for its respondent-count label.
node_gap = 70.0
x_positions = collect(1.0:n_wave)
half_width = 0.09
node_tops = Matrix{Float64}(undef, n_wave, n_cat)
node_bottoms = Matrix{Float64}(undef, n_wave, n_cat)
for w in 1:n_wave
y_cursor = sum(wave_counts[w]) + node_gap * (n_cat - 1)
for c in 1:n_cat
node_tops[w, c] = y_cursor
node_bottoms[w, c] = y_cursor - wave_counts[w][c]
y_cursor -= wave_counts[w][c] + node_gap
end
end
max_top = maximum(node_tops)
header_y = max_top + 65.0
# Plot — see default-style-guide.md "Visual Sizing Defaults" for the canvas + sizing values
title_str = "Political Opinion Survey · alluvial-opinion-flow · julia · makie · anyplot.ai"
title_default = 20
title_floor = 14
title_ratio = length(title_str) > 67 ? 67 / length(title_str) : 1.0
title_fontsize = max(title_floor, round(Int, title_default * title_ratio))
fig = Figure(
size = (1600, 900),
fontsize = 14,
backgroundcolor = PAGE_BG,
)
ax = Axis(
fig[1, 1];
title = title_str,
titlesize = title_fontsize,
titlecolor = INK,
backgroundcolor = PAGE_BG,
xticksvisible = false,
xticklabelsvisible = false,
yticksvisible = false,
yticklabelsvisible = false,
topspinevisible = false,
rightspinevisible = false,
leftspinevisible = false,
bottomspinevisible = false,
xgridvisible = false,
ygridvisible = false,
)
xlims!(ax, x_positions[1] - 0.85, x_positions[end] + 0.85)
ylims!(ax, -25.0, max_top + 110.0)
# Flows — smoothstep ribbons; opacity distinguishes stable respondents (same
# category both waves) from changers, per spec.
for w in 1:(n_wave - 1)
left_cursor = copy(node_tops[w, :])
right_cursor = copy(node_tops[w + 1, :])
for s in 1:n_cat, t in 1:n_cat
count = flow_counts[w][s, t]
count == 0 && continue
y0_top = left_cursor[s]
y0_bot = y0_top - count
left_cursor[s] = y0_bot
y1_top = right_cursor[t]
y1_bot = y1_top - count
right_cursor[t] = y1_bot
x0 = x_positions[w] + half_width
x1 = x_positions[w + 1] - half_width
xs = range(x0, x1; length = 40)
ts = range(0.0, 1.0; length = 40)
ease = @. ts^2 * (3 - 2 * ts)
top_curve = y0_top .+ (y1_top - y0_top) .* ease
bot_curve = y0_bot .+ (y1_bot - y0_bot) .* ease
flow_alpha = s == t ? 0.80 : 0.30
band!(ax, xs, bot_curve, top_curve; color = (IMPRINT_PALETTE[s], flow_alpha))
end
end
# Nodes — solid bars sized by respondent count, drawn above the flow ends.
for w in 1:n_wave, c in 1:n_cat
x0 = x_positions[w] - half_width
x1 = x_positions[w] + half_width
poly!(
ax,
Rect2f(x0, node_bottoms[w, c], x1 - x0, node_tops[w, c] - node_bottoms[w, c]);
color = IMPRINT_PALETTE[c],
strokewidth = 0,
)
text!(
ax, x_positions[w], node_tops[w, c] + 8.0;
text = string(wave_counts[w][c]),
align = (:center, :bottom),
fontsize = 12,
color = INK_SOFT,
)
end
# Column headers — one per wave, above the tallest column.
for w in 1:n_wave
text!(
ax, x_positions[w], header_y;
text = waves[w],
align = (:center, :bottom),
fontsize = 16,
color = INK,
font = :bold,
)
end
# Category names — labeled once on each outer column to avoid mid-chart clutter;
# the Imprint color already carries the identity across the interior waves.
for c in 1:n_cat
y_mid_left = (node_tops[1, c] + node_bottoms[1, c]) / 2
text!(
ax, x_positions[1] - half_width - 0.06, y_mid_left;
text = categories[c],
align = (:right, :center),
fontsize = 13,
color = INK,
)
y_mid_right = (node_tops[n_wave, c] + node_bottoms[n_wave, c]) / 2
text!(
ax, x_positions[n_wave] + half_width + 0.06, y_mid_right;
text = categories[c],
align = (:left, :center),
fontsize = 13,
color = INK,
)
end
# Save
save("plot-$(THEME).png", fig; px_per_unit = 2)
Part of Opinion Flow Diagram on anyplot.ai.