Chernoff faces visualize multivariate data by mapping each variable to a facial feature (eye size, mouth curvature, face width, nose length, etc.), transforming each observation into a unique cartoon face. This technique leverages humans' innate ability to recognize and distinguish faces, making it easier to identify patterns, clusters, and outliers across multiple dimensions simultaneously.
# anyplot.ai
# chernoff-basic: Chernoff Faces for Multivariate Data
# Library: makie 0.21.9 | Julia 1.11.9
# Quality: 96/100 | Created: 2026-09-02
using CairoMakie
using Colors
using Random
Random.seed!(42)
# --- Theme tokens -------------------------------------------------------------
const THEME = get(ENV, "ANYPLOT_THEME", "light")
const PAGE_BG = THEME == "light" ? colorant"#FAF8F1" : colorant"#1A1A17"
const ELEVATED_BG = THEME == "light" ? colorant"#FFFDF6" : colorant"#242420"
const INK = THEME == "light" ? colorant"#1A1A17" : colorant"#F0EFE8"
const INK_SOFT = THEME == "light" ? colorant"#4A4A44" : colorant"#B8B7B0"
const BRAND = colorant"#009E73" # Imprint palette position 1 -- ALWAYS first series
const ANYPLOT_AMBER = colorant"#DDCC77" # warning / caution -- flags the outlier patient
# --- Chernoff face recipe -------------------------------------------------------
# A custom Makie recipe: `chernoffface!` is a reusable, self-contained glyph
# (face outline + eyes/pupils + eyebrows + nose + mouth + label) whose shape is
# entirely driven by declarative attributes. This leans on Makie's recipe
# system (`@recipe`, attribute-linked sub-plots) rather than a generic
# poly!/lines!/text! loop.
@recipe(ChernoffFace, cx, cy) do scene
Attributes(
face_width = 0.5,
face_height = 0.5,
eye_size = 0.5,
eye_spacing = 0.5,
eyebrow_slant = 0.5,
nose_length = 0.5,
mouth_curve = 0.5,
mouth_width = 0.5,
rx_base = 0.55,
ry_base = 0.68,
facecolor = :white,
outlinecolor = :black,
outlinewidth = 3.0,
ink = :black,
ink_soft = :gray,
label = "",
labelcolor = :gray,
labelsize = 13.0,
)
end
function Makie.plot!(cf::ChernoffFace)
cx = cf[1][]
cy = cf[2][]
rx = cf.rx_base[] * (0.75 + 0.55 * cf.face_width[])
ry = cf.ry_base[] * (0.75 + 0.55 * cf.face_height[])
θ_face = range(0, 2π; length = 80)
face_pts = [Point2f(cx + rx * cos(t), cy + ry * sin(t)) for t in θ_face]
poly!(cf, face_pts; color = cf.facecolor, strokecolor = cf.outlinecolor,
strokewidth = cf.outlinewidth)
eye_r = 0.05 + 0.09 * cf.eye_size[]
eye_dx = rx * (0.30 + 0.24 * cf.eye_spacing[])
eye_y = cy + ry * 0.15
θ_eye = range(0, 2π; length = 40)
for side in (-1, 1)
ex = cx + side * eye_dx
eye_pts = [Point2f(ex + eye_r * cos(t), eye_y + eye_r * sin(t)) for t in θ_eye]
poly!(cf, eye_pts; color = cf.facecolor, strokecolor = cf.ink, strokewidth = 2)
pupil_pts = [Point2f(ex + 0.4 * eye_r * cos(t), eye_y + 0.4 * eye_r * sin(t)) for t in θ_eye]
poly!(cf, pupil_pts; color = cf.ink, strokewidth = 0)
end
brow_half_len = rx * 0.32
brow_y = eye_y + eye_r * 1.9
brow_slope = (0.5 - cf.eyebrow_slant[]) * 0.32 * ry
for (side, mirror) in ((-1, 1), (1, -1))
bx = cx + side * eye_dx
dy = mirror * brow_slope
lines!(cf, [Point2f(bx - brow_half_len, brow_y - dy), Point2f(bx + brow_half_len, brow_y + dy)];
color = cf.ink_soft, linewidth = 4)
end
nose_len = ry * (0.22 + 0.30 * cf.nose_length[])
nose_top = cy + ry * 0.02
lines!(cf, [Point2f(cx, nose_top), Point2f(cx, nose_top - nose_len)];
color = cf.ink_soft, linewidth = 2.5)
mouth_width = rx * (0.55 + 0.55 * cf.mouth_width[])
mouth_base_y = cy - ry * 0.42
mouth_a = (0.5 - cf.mouth_curve[]) * 0.9 * ry / max((mouth_width / 2)^2, 1e-6)
mouth_xs = range(-mouth_width / 2, mouth_width / 2; length = 30)
mouth_pts = [Point2f(cx + xv, mouth_base_y + mouth_a * xv^2) for xv in mouth_xs]
lines!(cf, mouth_pts; color = cf.ink, linewidth = 3.5)
text!(cf, cx, cy - ry - 0.16; text = cf.label, color = cf.labelcolor,
fontsize = cf.labelsize, align = (:center, :top))
cf
end
# --- Data: patient vital-sign profiles -----------------------------------------
n = 12
patient_ids = [string("P", lpad(i, 2, '0')) for i in 1:n]
resting_heart_rate = clamp.(72 .+ 12 .* randn(n), 50, 110) # bpm -> eye size
systolic_bp = clamp.(122 .+ 14 .* randn(n), 95, 165) # mmHg -> face width
cholesterol = clamp.(195 .+ 30 .* randn(n), 130, 280) # mg/dL -> eyebrow slant
bmi = clamp.(26 .+ 4 .* randn(n), 18, 38) # kg/m^2 -> face height
blood_glucose = clamp.(100 .+ 18 .* randn(n), 75, 160) # mg/dL -> mouth curvature
sleep_hours = clamp.(6.8 .+ 1.1 .* randn(n), 4.5, 9.0) # hours -> mouth width
respiratory_rate = clamp.(15 .+ 2.5 .* randn(n), 11, 22) # breaths/minute -> nose length
body_temperature = clamp.(98.2 .+ 0.6 .* randn(n), 96.8, 100.4) # deg F -> eye spacing
# Min-max normalize each variable to [0, 1] before mapping to a facial feature
eye_size_n = (resting_heart_rate .- minimum(resting_heart_rate)) ./ (maximum(resting_heart_rate) - minimum(resting_heart_rate))
face_width_n = (systolic_bp .- minimum(systolic_bp)) ./ (maximum(systolic_bp) - minimum(systolic_bp))
eyebrow_n = (cholesterol .- minimum(cholesterol)) ./ (maximum(cholesterol) - minimum(cholesterol))
face_height_n = (bmi .- minimum(bmi)) ./ (maximum(bmi) - minimum(bmi))
mouth_curve_n = (blood_glucose .- minimum(blood_glucose)) ./ (maximum(blood_glucose) - minimum(blood_glucose))
mouth_width_n = (sleep_hours .- minimum(sleep_hours)) ./ (maximum(sleep_hours) - minimum(sleep_hours))
nose_len_n = (respiratory_rate .- minimum(respiratory_rate)) ./ (maximum(respiratory_rate) - minimum(respiratory_rate))
eye_spacing_n = (body_temperature .- minimum(body_temperature)) ./ (maximum(body_temperature) - minimum(body_temperature))
# Flag the most extreme combined profile -- largest total deviation from the
# cohort midpoint (0.5) across all 8 normalized variables -- as a visual entry
# point into the comparison.
normalized = hcat(eye_size_n, face_width_n, eyebrow_n, face_height_n,
mouth_curve_n, mouth_width_n, nose_len_n, eye_spacing_n)
extremity = vec(sum((normalized .- 0.5) .^ 2; dims = 2))
outlier_idx = argmax(extremity)
# --- Grid layout: 4 columns x 3 rows -------------------------------------------
ncols, nrows = 4, 3
spacing_x, spacing_y = 2.0, 2.6
centers = Point2f[]
for i in 1:n
row = div(i - 1, ncols)
col = mod(i - 1, ncols)
push!(centers, Point2f(col * spacing_x, -row * spacing_y))
end
rx_max = 0.55 * 1.30
ry_max = 0.68 * 1.30
# --- Figure -------------------------------------------------------------------
fig = Figure(
resolution = (1200, 1200),
fontsize = 14,
backgroundcolor = PAGE_BG,
)
ax = Axis(
fig[1, 1];
title = "chernoff-basic · julia · makie · anyplot.ai",
titlesize = 26,
titlecolor = INK,
backgroundcolor = PAGE_BG,
aspect = DataAspect(),
)
hidedecorations!(ax)
hidespines!(ax)
xlims!(ax, -rx_max - 0.3, (ncols - 1) * spacing_x + rx_max + 0.3)
ylims!(ax, -(nrows - 1) * spacing_y - ry_max - 0.55, ry_max + 0.3)
# --- Draw one Chernoff face per patient via the custom recipe -------------------
for i in 1:n
is_outlier = i == outlier_idx
chernoffface!(ax, centers[i][1], centers[i][2];
face_width = face_width_n[i],
face_height = face_height_n[i],
eye_size = eye_size_n[i],
eye_spacing = eye_spacing_n[i],
eyebrow_slant = eyebrow_n[i],
nose_length = nose_len_n[i],
mouth_curve = mouth_curve_n[i],
mouth_width = mouth_width_n[i],
facecolor = ELEVATED_BG,
outlinecolor = is_outlier ? ANYPLOT_AMBER : BRAND,
outlinewidth = is_outlier ? 5.0 : 3.0,
ink = INK,
ink_soft = INK_SOFT,
label = patient_ids[i],
labelcolor = INK_SOFT,
labelsize = 13,
)
end
# --- Save -----------------------------------------------------------------------
save("plot-$(THEME).png", fig; px_per_unit = 2)
Runnable source as JSON, for any HTTP client: https://api.anyplot.ai/specs/chernoff-basic/makie/code. Any spec id and library id listed in llms-full.txt fit the same URL shape; every URL below is complete and callable.
{
"spec_id": "chernoff-basic",
"language": "julia",
"library": "makie",
"page": "https://anyplot.ai/chernoff-basic/julia/makie",
"hub": "https://anyplot.ai/chernoff-basic",
"code_json": "https://api.anyplot.ai/specs/chernoff-basic/makie/code",
"spec_json": "https://api.anyplot.ai/specs/chernoff-basic",
"render_light_png": "https://storage.googleapis.com/anyplot-images/plots/chernoff-basic/julia/makie/plot-light.png",
"render_dark_png": "https://storage.googleapis.com/anyplot-images/plots/chernoff-basic/julia/makie/plot-dark.png",
"quality_score": 96.0,
"license": "MIT",
"guide": "https://anyplot.ai/llms.txt"
}Part of Chernoff Faces for Multivariate Data on anyplot.ai.