Mosaic Plot for Categorical Association Analysis — plotnine

A mosaic plot visualizes contingency tables by dividing a rectangular area into smaller rectangles whose areas are proportional to cell frequencies. This statistical visualization technique effectively shows relationships and associations between two or more categorical variables, making it easy to identify patterns, dependencies, and deviations from expected frequencies in cross-tabulated data.

Mosaic Plot for Categorical Association Analysis rendered with plotnine

Python source (plotnine)

""" anyplot.ai
mosaic-categorical: Mosaic Plot for Categorical Association Analysis
Library: plotnine 0.15.4 | Python 3.13.13
Quality: 89/100 | Updated: 2026-05-19
"""

import os

import pandas as pd
from plotnine import (
    aes,
    element_blank,
    element_line,
    element_rect,
    element_text,
    geom_rect,
    geom_text,
    ggplot,
    labs,
    scale_fill_manual,
    scale_x_continuous,
    scale_y_continuous,
    theme,
    theme_minimal,
)


# Theme tokens
THEME = os.getenv("ANYPLOT_THEME", "light")
PAGE_BG = "#FAF8F1" if THEME == "light" else "#1A1A17"
ELEVATED_BG = "#FFFDF6" if THEME == "light" else "#242420"
INK = "#1A1A17" if THEME == "light" else "#F0EFE8"
INK_SOFT = "#4A4A44" if THEME == "light" else "#B8B7B0"
INK_MUTED = "#6B6A63" if THEME == "light" else "#A8A79F"

IMPRINT = ["#009E73", "#C475FD", "#4467A3", "#BD8233", "#AE3030", "#2ABCCD", "#954477"]

# Data - Titanic survival data by passenger class
data = {
    "class": ["First", "First", "Second", "Second", "Third", "Third"],
    "survival": ["Survived", "Did Not Survive", "Survived", "Did Not Survive", "Survived", "Did Not Survive"],
    "count": [203, 122, 118, 167, 178, 528],
}
df = pd.DataFrame(data)
survival_order = ["Survived", "Did Not Survive"]

# Calculate proportions for mosaic geometry
total = df["count"].sum()
class_totals = df.groupby("class")["count"].sum()
class_order = ["First", "Second", "Third"]
widths = {c: class_totals[c] / total for c in class_order}

# Build rectangle data for each cell — widths ∝ class size, heights ∝ conditional proportion
gap = 0.02
rects = []
x_pos = 0
x_centers = {}
survival_rates = {}

for cls in class_order:
    class_data = df[df["class"] == cls]
    class_total = class_data["count"].sum()
    width = widths[cls] - gap
    x_centers[cls] = x_pos + width / 2
    survived_n = class_data[class_data["survival"] == "Survived"]["count"].values[0]
    survival_rates[cls] = survived_n / class_total

    y_pos = 0
    for surv in survival_order:
        row = class_data[class_data["survival"] == surv].iloc[0]
        height = (row["count"] / class_total) * (1 - gap)
        rects.append(
            {
                "xmin": x_pos,
                "xmax": x_pos + width,
                "ymin": y_pos,
                "ymax": y_pos + height,
                "survival": row["survival"],
                "count": row["count"],
                "x_center": x_pos + width / 2,
                "y_center": y_pos + height / 2,
            }
        )
        y_pos += height + gap / 2

    x_pos += widths[cls]

rect_df = pd.DataFrame(rects)

# Class name labels and per-class survival rate annotations below the mosaic
class_labels = pd.DataFrame({"x": [x_centers[c] for c in class_order], "y": [-0.055] * 3, "label": class_order})
rate_labels = pd.DataFrame(
    {
        "x": [x_centers[c] for c in class_order],
        "y": [-0.115] * 3,
        "label": [f"{survival_rates[c]:.0%} survived" for c in class_order],
    }
)

# Okabe-Ito: green → Survived (first series), vermillion → Did Not Survive
colors = {"Survived": IMPRINT[0], "Did Not Survive": IMPRINT[1]}

# scale_x_continuous / scale_y_continuous give explicit domain + expansion control —
# avoids clipping the below-axis class and rate labels
plot = (
    ggplot(rect_df)
    + geom_rect(aes(xmin="xmin", xmax="xmax", ymin="ymin", ymax="ymax", fill="survival"), color=PAGE_BG, size=1.5)
    + geom_text(aes(x="x_center", y="y_center", label="count"), color="white", size=14, fontweight="bold")
    + geom_text(data=class_labels, mapping=aes(x="x", y="y", label="label"), size=14, color=INK, fontweight="bold")
    + geom_text(data=rate_labels, mapping=aes(x="x", y="y", label="label"), size=12, color=INK_SOFT)
    + scale_fill_manual(values=colors, breaks=["Survived", "Did Not Survive"])
    + scale_x_continuous(expand=(0, 0.01), limits=(0, 1.01))
    + scale_y_continuous(expand=(0, 0), limits=(-0.17, 1.04))
    + labs(
        title="mosaic-categorical · python · plotnine · anyplot.ai",
        subtitle="First-class survival rate (62%) was 2.5× higher than Third-class (25%)",
        y="Conditional Survival Proportion",
        fill="Outcome",
    )
    + theme_minimal()
    + theme(
        figure_size=(16, 9),
        plot_background=element_rect(fill=PAGE_BG, color=PAGE_BG),
        panel_background=element_rect(fill=PAGE_BG),
        panel_border=element_blank(),
        panel_grid=element_blank(),
        axis_line_x=element_line(color=INK_SOFT, size=0.5),
        axis_line_y=element_line(color=INK_SOFT, size=0.5),
        plot_title=element_text(size=24, ha="center", color=INK, fontweight="bold"),
        plot_subtitle=element_text(size=17, ha="center", color=INK_SOFT),
        axis_title_x=element_blank(),
        axis_title_y=element_text(size=20, color=INK),
        axis_text_x=element_blank(),
        axis_ticks_major_x=element_blank(),
        axis_text_y=element_text(size=16, color=INK_SOFT),
        legend_title=element_text(size=18, color=INK),
        legend_text=element_text(size=16, color=INK_SOFT),
        legend_background=element_rect(fill=ELEVATED_BG, color=INK_SOFT),
        legend_position="right",
    )
)

plot.save(f"plot-{THEME}.png", dpi=300)

Part of Mosaic Plot for Categorical Association Analysis on anyplot.ai.

Other implementations