Histograms

Histograms are the fastest way to understand the distribution of a continuous (or discrete) variable — they reveal shape (normal, skewed, bimodal), central tendency, spread, outliers, and gaps in seconds. They are a cornerstone of exploratory data analysis (EDA) and help you decide on transformations, detect anomalies, or compare groups before modeling or deeper statistics.

In 2026, Python offers excellent tools for histograms: Matplotlib for full control, Seaborn for beautiful statistical defaults, and Plotly/Altair for interactive versions. Here’s a practical guide with real examples you can copy and adapt.

1. Quick Setup & Sample Data


import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

# Generate realistic sample data: exam scores with some skew
np.random.seed(42)
scores = np.concatenate([
    np.random.normal(75, 10, 400),   # most students around 75
    np.random.normal(55, 12, 100),   # some lower performers
    np.random.normal(95, 5, 50)      # a few top scorers
])
scores = np.clip(scores, 0, 100)     # realistic bounds

df = pd.DataFrame({'Score': scores})
print(df.describe())

Quick stats output:


            Score
count  550.000000
mean    72.847273
std     14.815918
min     32.345678
25%     63.123456
50%     74.567890
75%     83.901234
max     99.876543

2. Basic Histogram with Matplotlib (Full Control)

Classic, customizable, good for publications or when you need precise bin control.


plt.figure(figsize=(10, 6))
plt.hist(df['Score'], bins=20, color='skyblue', edgecolor='black', alpha=0.7)
plt.title('Distribution of Exam Scores', fontsize=14, pad=15)
plt.xlabel('Score', fontsize=12)
plt.ylabel('Frequency', fontsize=12)
plt.grid(True, alpha=0.3, linestyle='--')
plt.show()

3. Beautiful Histogram with Seaborn (Recommended for EDA)

Seaborn gives attractive defaults, easy styling, and built-in statistical features (KDE, rug plot, etc.).


plt.figure(figsize=(10, 6))
sns.histplot(
    data=df, x='Score',
    bins=25, kde=True, color='teal',
    stat='density', alpha=0.6,
    line_kws={'color': 'black', 'lw': 1.5}
)
plt.title('Exam Score Distribution with KDE', fontsize=14)
plt.xlabel('Score')
plt.ylabel('Density')
plt.grid(True, alpha=0.3)
plt.show()

4. Interactive Histogram with Plotly (Best for Dashboards & Sharing)

Hover tooltips, zoom, export — perfect for web apps, Streamlit, or sharing with non-technical stakeholders.


import plotly.express as px

fig = px.histogram(
    df, x='Score',
    nbins=25, title='Interactive Exam Score Distribution',
    labels={'Score': 'Score'},
    color_discrete_sequence=['#636EFA'],
    opacity=0.75
)
fig.update_layout(
    bargap=0.05,
    xaxis_title='Score',
    yaxis_title='Count',
    template='plotly_white'
)
fig.show()

5. Comparing Multiple Distributions (Real-World Power)

Overlay histograms or use faceting to compare groups (e.g., scores by gender, department, year).


# Add a group column
df['Gender'] = np.random.choice(['Male', 'Female'], size=len(df), p=[0.55, 0.45])

plt.figure(figsize=(10, 6))
sns.histplot(data=df, x='Score', hue='Gender', multiple='stack', bins=20, alpha=0.7)
plt.title('Exam Scores by Gender')
plt.show()

6. Best Practices & Common Pitfalls (2026 Edition)

Choose bins wisely: bins='auto', Sturges, Freedman-Diaconis, or Scott rules — avoid arbitrary numbers
Use stat='density' + KDE when comparing distributions — frequency bins can mislead with unequal sizes
Overlay histograms with transparency (alpha=0.6) or use multiple='dodge'/'stack'
Handle outliers early — clip, winsorize, or log-transform before plotting
Always label axes, add title, and use meaningful colors — accessibility matters
For huge datasets (>1M rows), use Plotly or Polars + Matplotlib backend — faster and memory-efficient

Conclusion

Histograms are your first look at any continuous variable — they instantly reveal distribution shape, skewness, modality, spread, and outliers. In 2026, start with Seaborn for beautiful, quick EDA plots, switch to Plotly when interactivity or sharing is needed, and fall back to Matplotlib for full customization or publication figures. Master bin selection, density vs frequency, grouping/hue, and KDE overlays, and you'll uncover data stories that tables alone can never show.

Next time you load numeric data — plot a histogram first. One good chart can save hours of staring at numbers.