Seaborn

Seaborn is a statistical data visualization library built on Matplotlib. It provides beautiful defaults, works directly with Pandas DataFrames, and produces complex statistical plots in one line.

Seaborn vs Matplotlib

Setup and Themes

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

sns.set_theme(style="whitegrid", palette="deep", font_scale=1.05)
# styles:   darkgrid, whitegrid, dark, white, ticks
# palettes: deep, muted, pastel, bright, dark, colorblind, viridis, Set2

tips = sns.load_dataset("tips")
iris = sns.load_dataset("iris")
print(tips.head())

Categories of Seaborn Plots

1. Distribution Plots

fig, axes = plt.subplots(2, 2, figsize=(13, 8))

sns.histplot(data=tips, x="total_bill", bins=25, kde=True, ax=axes[0,0], color="#168B99")
axes[0,0].set_title("histplot — histogram + KDE")

sns.kdeplot(data=tips, x="total_bill", hue="time", fill=True, alpha=0.4, ax=axes[0,1])
axes[0,1].set_title("kdeplot — smooth density by group")

sns.ecdfplot(data=tips, x="total_bill", hue="sex", ax=axes[1,0])
axes[1,0].set_title("ecdfplot — cumulative distribution")

sns.histplot(data=tips, x="total_bill", y="tip", bins=20, cbar=True, ax=axes[1,1])
axes[1,1].set_title("histplot 2-D — bivariate density")

plt.tight_layout(); plt.show()

2. Categorical Plots

fig, axes = plt.subplots(2, 3, figsize=(17, 9))

sns.countplot(data=tips, x="day", ax=axes[0,0], palette="Set2")
axes[0,0].set_title("countplot — frequency of each category")

sns.barplot(data=tips, x="day", y="total_bill", ax=axes[0,1], palette="Set2",
            errorbar=("ci", 95))
axes[0,1].set_title("barplot — mean with 95% CI")

sns.boxplot(data=tips, x="day", y="total_bill", hue="sex", ax=axes[0,2], palette="Set1")
axes[0,2].set_title("boxplot — five-number summary")

sns.violinplot(data=tips, x="day", y="total_bill", ax=axes[1,0], palette="Set3",
               inner="quartile")
axes[1,0].set_title("violinplot — box plot + distribution shape")

sns.stripplot(data=tips, x="day", y="total_bill", ax=axes[1,1], alpha=0.5, jitter=0.25)
axes[1,1].set_title("stripplot — every raw point")

sns.pointplot(data=tips, x="day", y="total_bill", hue="time", ax=axes[1,2],
              dodge=True, errorbar="se")
axes[1,2].set_title("pointplot — means connected, shows interaction")

plt.tight_layout(); plt.show()

barplot shows the MEAN of a numeric variable per category; countplot shows the COUNT of rows per category. Confusing the two is a common mistake.

3. Relational Plots

fig, axes = plt.subplots(1, 3, figsize=(17, 4.8))

sns.scatterplot(data=tips, x="total_bill", y="tip", hue="time",
                style="smoker", size="size", sizes=(30, 250), alpha=0.7, ax=axes[0])
axes[0].set_title("scatterplot — 5 variables at once")

flights = sns.load_dataset("flights")
sns.lineplot(data=flights, x="year", y="passengers", ax=axes[1],
             errorbar=("ci", 95), marker="o")
axes[1].set_title("lineplot — trend with confidence band")

sns.lineplot(data=flights, x="month", y="passengers", hue="year",
             palette="viridis", legend=False, ax=axes[2])
axes[2].set_title("lineplot — seasonality by year")
axes[2].tick_params(axis="x", rotation=45)

plt.tight_layout(); plt.show()

4. Regression Plots

fig, axes = plt.subplots(1, 3, figsize=(17, 4.8))

sns.regplot(data=tips, x="total_bill", y="tip", ax=axes[0],
            scatter_kws={"alpha": 0.5}, line_kws={"color": "#ef4444"})
axes[0].set_title("regplot — linear fit with 95% CI")

sns.regplot(data=tips, x="total_bill", y="tip", order=2, ax=axes[1],
            scatter_kws={"alpha": 0.4}, line_kws={"color": "#10b981"})
axes[1].set_title("regplot order=2 — polynomial fit")

sns.residplot(data=tips, x="total_bill", y="tip", ax=axes[2], color="#6366f1")
axes[2].set_title("residplot — diagnostics (want a random cloud)")

plt.tight_layout(); plt.show()

# lmplot = regplot + FacetGrid (creates its own figure)
sns.lmplot(data=tips, x="total_bill", y="tip", col="time", hue="smoker",
           height=4, aspect=1.1)
plt.show()

5. Matrix Plots

fig, axes = plt.subplots(1, 2, figsize=(14, 5.5))

corr = tips[["total_bill", "tip", "size"]].corr()
sns.heatmap(corr, annot=True, fmt=".3f", cmap="coolwarm", center=0,
            square=True, linewidths=1, cbar_kws={"shrink": 0.8}, ax=axes[0])
axes[0].set_title("heatmap — correlation matrix")

pivot = flights.pivot(index="month", columns="year", values="passengers")
sns.heatmap(pivot, cmap="YlOrRd", ax=axes[1], cbar_kws={"label": "Passengers"})
axes[1].set_title("heatmap — time x category matrix")

plt.tight_layout(); plt.show()

# clustermap — heatmap with hierarchical clustering (Unit 3) applied to rows/columns
sns.clustermap(iris.drop(columns="species").corr(), annot=True,
               cmap="coolwarm", center=0, figsize=(6.5, 6.5))
plt.show()

# Masking the upper triangle — the standard correlation heatmap presentation
numeric = iris.select_dtypes(include=np.number)
corr = numeric.corr()
mask = np.triu(np.ones_like(corr, dtype=bool))

plt.figure(figsize=(7.5, 6))
sns.heatmap(corr, mask=mask, annot=True, fmt=".2f", cmap="coolwarm",
            center=0, square=True, linewidths=1, vmin=-1, vmax=1)
plt.title("Iris — Correlation (lower triangle only)", fontweight="bold")
plt.tight_layout(); plt.show()

6. Multi-Plot Grids

# pairplot — every numeric pair, coloured by a category
sns.pairplot(iris, hue="species", diag_kind="kde", height=2.2,
             plot_kws={"alpha": 0.7, "s": 35}, corner=False)
plt.suptitle("Iris — Pair Plot", y=1.01, fontweight="bold")
plt.show()

# jointplot — a bivariate plot with marginal distributions
sns.jointplot(data=tips, x="total_bill", y="tip", kind="reg", height=6)
plt.show()
sns.jointplot(data=tips, x="total_bill", y="tip", kind="hex", height=6)
plt.show()
sns.jointplot(data=tips, x="total_bill", y="tip", hue="time", kind="kde", height=6)
plt.show()

# FacetGrid — the same plot repeated across subsets of the data
g = sns.FacetGrid(tips, col="time", row="smoker", height=3.4, aspect=1.25,
                  margin_titles=True)
g.map_dataframe(sns.scatterplot, x="total_bill", y="tip", alpha=0.7)
g.set_axis_labels("Total Bill", "Tip")
g.figure.suptitle("Tip vs Bill across Time and Smoker Status", y=1.03, fontweight="bold")
plt.show()

# The figure-level equivalents do this automatically
sns.catplot(data=tips, x="day", y="total_bill", hue="sex",
            col="time", kind="box", height=4, aspect=1.1)
plt.show()

sns.displot(data=tips, x="total_bill", col="day", kde=True,
            col_wrap=2, height=3.2)
plt.show()

Axes-Level vs Figure-Level Functions

Palettes and Styling

# Built-in palettes
sns.color_palette("deep")          # default qualitative
sns.color_palette("viridis", 8)    # sequential (ordered data)
sns.color_palette("coolwarm", 8)   # diverging (has a meaningful midpoint)
sns.color_palette("colorblind")    # accessible — use this for public reports

# Custom palette
custom = ["#168B99", "#10b981", "#f59e0b", "#6366f1", "#ef4444"]
sns.set_palette(custom)

fig, ax = plt.subplots(figsize=(9, 5))
sns.barplot(data=tips, x="day", y="total_bill", hue="sex",
            palette=["#168B99", "#f59e0b"], ax=ax, errorbar=None)
ax.set_title("Average Bill by Day and Sex", fontsize=14, fontweight="bold")
ax.set_xlabel("Day of Week"); ax.set_ylabel("Average Total Bill (Rs)")
sns.despine()                       # remove the top and right spines
ax.legend(title="Sex", frameon=False)
plt.tight_layout(); plt.show()

A Complete EDA Figure

sns.set_theme(style="whitegrid")
fig = plt.figure(figsize=(16, 11))
gs = fig.add_gridspec(3, 3, hspace=0.4, wspace=0.3)

ax1 = fig.add_subplot(gs[0, 0])
sns.histplot(data=tips, x="total_bill", kde=True, ax=ax1, color="#168B99")
ax1.set_title("Bill Distribution")

ax2 = fig.add_subplot(gs[0, 1])
sns.boxplot(data=tips, x="day", y="total_bill", ax=ax2, palette="Set2")
ax2.set_title("Bill by Day")

ax3 = fig.add_subplot(gs[0, 2])
sns.countplot(data=tips, x="day", hue="time", ax=ax3, palette="Set1")
ax3.set_title("Orders by Day and Time")

ax4 = fig.add_subplot(gs[1, :2])
sns.scatterplot(data=tips, x="total_bill", y="tip", hue="time", size="size",
                sizes=(25, 220), alpha=0.7, ax=ax4)
sns.regplot(data=tips, x="total_bill", y="tip", scatter=False,
            line_kws={"color": "#ef4444", "linewidth": 2}, ax=ax4)
ax4.set_title("Tip vs Bill with Trend Line")

ax5 = fig.add_subplot(gs[1, 2])
sns.heatmap(tips[["total_bill","tip","size"]].corr(), annot=True, fmt=".2f",
            cmap="coolwarm", center=0, square=True, ax=ax5, cbar=False)
ax5.set_title("Correlations")

ax6 = fig.add_subplot(gs[2, 0])
sns.violinplot(data=tips, x="sex", y="tip", hue="smoker", split=True,
               ax=ax6, palette="Set2")
ax6.set_title("Tip by Sex and Smoker")

ax7 = fig.add_subplot(gs[2, 1])
tips_pct = tips.assign(tip_pct=lambda d: d["tip"] / d["total_bill"] * 100)
sns.barplot(data=tips_pct, x="size", y="tip_pct", ax=ax7, palette="viridis",
            errorbar=None)
ax7.set_title("Tip % by Party Size")

ax8 = fig.add_subplot(gs[2, 2])
sns.kdeplot(data=tips_pct, x="tip_pct", hue="time", fill=True, alpha=0.4, ax=ax8)
ax8.set_title("Tip % Distribution")

fig.suptitle("Restaurant Tips — Complete Exploratory Analysis",
             fontsize=17, fontweight="bold", y=0.995)
plt.show()

Quick Reference — Which Seaborn Function

With NumPy, Pandas, Matplotlib and Seaborn, you can analyse any dataset that fits in memory. The remaining lessons cover what happens when it doesn't.