Cross-Tabulation

Descriptive Statistics

Unlock the Hidden Relationships in Your Categorical Data

Cross-tabulation reveals how two categorical variables interact — turning raw counts into actionable insights about association and dependence.

Discover patterns — See how categories of one variable distribute across another
Compare groups — Use row, column, and total percentages to uncover disparities
Test significance — Apply chi-square testing to determine if associations are real or due to chance
Visualize findings — Transform tables into heatmaps that tell a compelling story

Every contingency table is a window into the structure of your data.

What is Cross-Tabulation?

Definition

Cross-tabulation (cross-tab or crosstab) displays the joint distribution of two or more categorical variables in a matrix format.

import numpy as np
import pandas as pd
from scipy import stats

np.random.seed(42)

# Sample data
n = 200
data = pd.DataFrame({
    'Gender': np.random.choice(['Male', 'Female'], n),
    'Preference': np.random.choice(['Tea', 'Coffee', 'Juice'], n),
    'Age_Group': np.random.choice(['18-30', '31-50', '51+'], n)
})

# Basic cross-tabulation
ct = pd.crosstab(data['Gender'], data['Preference'])
print("Cross-Tabulation: Gender vs Preference")
print(ct)

Row and Column Percentages

# Row percentages (normalize by row)
ct_row = pd.crosstab(data['Gender'], data['Preference'], normalize='index') * 100
print("Row Percentages (%):\n", ct_row.round(1))

# Column percentages (normalize by column)
ct_col = pd.crosstab(data['Gender'], data['Preference'], normalize='columns') * 100
print("\nColumn Percentages (%):\n", ct_col.round(1))

# Overall percentages
ct_all = pd.crosstab(data['Gender'], data['Preference'], normalize='all') * 100
print("\nOverall Percentages (%):\n", ct_all.round(1))

Three-Way Cross-Tabulation

# Three-way crosstab
ct_3way = pd.crosstab(
    [data['Gender'], data['Age_Group']],
    data['Preference'],
    margins=True
)
print("Three-way Cross-Tabulation:")
print(ct_3way)

Chi-Square Test of Independence

ct_table = pd.crosstab(data['Gender'], data['Preference'])
chi2, p_value, dof, expected = stats.chi2_contingency(ct_table)

print(f"Chi-square = {chi2:.4f}")
print(f"p-value    = {p_value:.4f}")
print(f"Degrees of freedom = {dof}")
print(f"\nExpected frequencies:\n{pd.DataFrame(expected, index=ct_table.index, columns=ct_table.columns).round(1)}")

Heatmap Visualization

import matplotlib.pyplot as plt
import seaborn as sns

fig, axes = plt.subplots(1, 2, figsize=(12, 5))

sns.heatmap(ct, annot=True, fmt='d', cmap='Blues', ax=axes[0])
axes[0].set_title('Counts')

sns.heatmap(ct_row, annot=True, fmt='.1f', cmap='Blues', ax=axes[1])
axes[1].set_title('Row Percentages (%)')

plt.tight_layout()
plt.savefig('cross-tabulation.png', dpi=150)
plt.show()

Cross-Tabulation in Machine Learning

ML Application	Cross-Tab Usage	Why
EDA	Explore feature relationships	Understand data structure
NLP	Word vs category counts	Feature engineering
A/B testing	Treatment vs outcome	Quick significance check
Data validation	Expected vs actual counts	Detect data issues

import numpy as np
import pandas as pd
from scipy.stats import chi2_contingency

np.random.seed(42)
n = 500
data = pd.DataFrame({
    'predicted': np.random.choice(['cat_a', 'cat_b', 'cat_c'], n),
    'actual': np.random.choice(['cat_a', 'cat_b', 'cat_c'], n)
})
ct = pd.crosstab(data['predicted'], data['actual'], margins=True)
print("Cross-tabulation (confusion matrix style):")
print(ct)

chi2, p, dof, _ = chi2_contingency(pd.crosstab(data['predicted'], data['actual']))
print(f"\nChi-square test: χ² = {chi2:.3f}, p = {p:.4f}")

Key Takeaways

Cross-tabs display joint distributions of two or more categorical variables in a matrix format.

Use row percentages to compare distributions across groups, and column percentages to compare group compositions within categories.

pandas crosstab() is the primary tool — it supports margins, normalization, and multi-level indexing for flexible analysis.

The chi-square test determines whether an observed association between variables is statistically significant or due to random chance.

Cross-tabulation is the Swiss Army knife of categorical data analysis — simple to construct, yet powerful enough to reveal the structure hidden in your data.

Cross-Tabulation — Analyzing Relationships Between Categorical Variables