Master Data Distribution Visualization with Seaborn: Histograms to Violin Plots

Effective data analysis often begins with understanding the distribution of a dataset, and seaborn offers a rich set of visual tools to explore these properties. The following sections demonstrate how to create and customize various distribution plots using the penguins dataset.

Univariate Histogram

A histogram bins data points and displays the frequency of each bin as bars. Different bin counts or widths can dramatically affect the visual result.

import seaborn as sns
data = sns.load_dataset("penguins", data_home="./data/seaborn-data")
# Different bin specifications
sns.displot(data, x='flipper_length_mm', bins=20)
sns.displot(data, x='flipper_length_mm', bins=50)
sns.displot(data, x='flipper_length_mm', binwidth=5)

Histogram Normalization

Normalization makes histograms comparable across datasets. Two common approaches are density normalization (area = 1) and probability normalization (sum of bar heights = 1).

# Density normalization
sns.displot(data, x='flipper_length_mm', bins=20, stat='density')
# Probability normalization
sns.displot(data, x='flipper_length_mm', bins=20, stat='probability')

Conditional Histogram

By setting the hue parameter, you can compare the distribution of a variable across categories such as species, sex, or island.

sns.displot(data, x="flipper_length_mm", hue='species')
sns.displot(data, x="flipper_length_mm", hue='sex')
sns.displot(data, x="flipper_length_mm", hue='island')

Normalization in Conditional Histograms

Setting common_norm=False normalizes each condition independently, highlighting differences in shape rather than scale.

sns.displot(data, x="flipper_length_mm", hue='species')
sns.displot(data, x="flipper_length_mm", hue='species', stat='probability')
sns.displot(data, x="flipper_length_mm", hue='species', stat='probability', common_norm=False)

Kernel Density Estimation (KDE)

KDE provides a smooth estimate of the underlying probability density. Bandwidth controls the smoothness; smaller values reveal more detail but can introduce noise.

sns.displot(data, x="flipper_length_mm", kind="kde", bw_adjust=1)
sns.displot(data, x="flipper_length_mm", kind="kde", bw_adjust=0.25)
sns.displot(data, x="flipper_length_mm", kind="kde", bw_adjust=1.5)

Combining Histogram and KDE

Overlaying a KDE curve on a histogram gives both discrete and continuous perspectives.

sns.displot(data, x="flipper_length_mm", kde=True)
sns.displot(data, x="flipper_length_mm", hue="species", kde=True)

Boxplot

Boxplots summarize key statistics (median, quartiles, IQR, outliers) and are useful for quick comparisons across categories.

sns.boxplot(data=data, x="flipper_length_mm", y='species', showmeans=True,
            meanprops={"marker":"o", "markerfacecolor":"red", "markeredgecolor":"black", "markersize":5})

sns.boxplot(data=data, x="flipper_length_mm", y='species', hue='sex', showmeans=True,
            meanprops={"marker":"o", "markerfacecolor":"red", "markeredgecolor":"black", "markersize":5})

sns.boxplot(data=data, x="flipper_length_mm", showmeans=True,
            meanprops={"marker":"o", "markerfacecolor":"red", "markeredgecolor":"black", "markersize":5})

Violin Plot

Violin plots combine a boxplot with a KDE curve, displaying the full distribution shape and individual sample points.

sns.violinplot(data=data, x="flipper_length_mm", y='species', inner="stick")
sns.violinplot(data=data, x="flipper_length_mm", y='species', hue='sex', split=True, inner="stick")
sns.violinplot(data=data, x="flipper_length_mm", inner="stick")

Empirical Cumulative Distribution Function (ECDF)

ECDF plots the cumulative proportion of observations without requiring binning or bandwidth selection.

sns.displot(data, x="flipper_length_mm", kind="ecdf")
sns.displot(data, x="flipper_length_mm", hue="species", kind="ecdf")

Bivariate Histogram and KDE

Two‑dimensional histograms (heatmaps) and bivariate KDEs reveal joint distributions. The multiple="stack" option can be used for stacked conditional visualizations.

sns.displot(data, x="flipper_length_mm", y="species", cbar=True)
sns.displot(data, x="bill_length_mm", y="bill_depth_mm", hue="species")
sns.displot(data, x="bill_length_mm", y="bill_depth_mm", hue="species", kind="kde")

Joint Plot and JointGrid

Joint plots combine scatter (or histogram) visualizations of two variables with marginal distributions. JointGrid allows custom combinations of plot types.

# Simple joint scatter with marginal histograms
sns.jointplot(data=data, x="bill_length_mm", y="bill_depth_mm")

# Joint KDE with marginal KDEs
sns.jointplot(data=data, x="bill_length_mm", y="bill_depth_mm", hue="species", kind="kde")

# Custom JointGrid examples
g = sns.JointGrid(data=data, x="bill_length_mm", y="bill_depth_mm")
g.plot_joint(sns.histplot)
g.plot_marginals(sns.kdeplot)

g = sns.JointGrid(data=data, x="bill_length_mm", y="bill_depth_mm")
g.plot_joint(sns.histplot)
g.plot_marginals(sns.boxplot)

g = sns.JointGrid(data=data, x="bill_length_mm", y="bill_depth_mm")
g.plot_joint(sns.scatterplot)
g.plot_marginals(sns.boxplot)

These seaborn visualizations provide a comprehensive toolbox for exploring and presenting data distributions, from simple univariate histograms to complex joint and conditional plots.