Data Visualisation with Matplotlib and Seaborn

Why Visualise?

A table of 10,000 numbers tells you almost nothing at a glance. A single chart can reveal a trend, an outlier or a relationship in seconds. Data scientists use charts for two different jobs:

1. Exploration: quick, rough charts to understand the data for yourself
2. Communication: polished charts that convince a manager at Zenith Bank or a product lead at Spotify to act

Python has two libraries you must know. Matplotlib is the low-level foundation that gives you full control. Seaborn is built on top of it and produces attractive statistical charts in one line.

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

sns.set_theme(style="whitegrid")

Matplotlib: The Figure and the Axes

Every Matplotlib chart has a Figure (the whole canvas) and one or more Axes (the individual plots). The object-oriented style below is the professional habit, because it scales to multi-panel figures.

months = ["Jan", "Feb", "Mar", "Apr", "May", "Jun"]
revenue = [42, 45, 51, 49, 58, 63]   # N million

fig, ax = plt.subplots(figsize=(8, 4))
ax.plot(months, revenue, marker="o", color="#EC4899")
ax.set_title("Monthly Revenue, H1 2025")
ax.set_xlabel("Month")
ax.set_ylabel("Revenue (N million)")
plt.show()

Always label the title, the axes and the units. A chart without units is a chart nobody can trust.

The Core Chart Types

Picking the right chart is a skill in its own right. The lab below trains you to match a question to a chart type.

Seaborn: Statistical Charts in One Line

Seaborn works directly with DataFrames. You pass the DataFrame and column names, and it handles grouping, colours and legends.

np.random.seed(0)
df = pd.DataFrame({
    "bank": np.random.choice(["Access", "GTBank", "Zenith", "UBA"], 400),
    "age": np.random.randint(18, 65, 400),
})
df["monthly_spend"] = df["age"] * 2500 + np.random.normal(0, 25000, 400)

# Distribution of one variable
sns.histplot(data=df, x="monthly_spend", bins=30, kde=True)
plt.show()

# Compare distributions across groups
sns.boxplot(data=df, x="bank", y="monthly_spend")
plt.show()

# Relationship between two variables, coloured by a category
sns.scatterplot(data=df, x="age", y="monthly_spend", hue="bank", alpha=0.7)
plt.show()

Multiple Charts in One Figure

Dashboards and reports often place charts side by side:

fig, axes = plt.subplots(1, 2, figsize=(12, 4))
sns.histplot(data=df, x="age", ax=axes[0])
axes[0].set_title("Customer Age Distribution")
sns.barplot(data=df, x="bank", y="monthly_spend", ax=axes[1])
axes[1].set_title("Average Monthly Spend by Bank")
plt.tight_layout()
plt.show()

Correlation Heatmaps

A heatmap of a correlation matrix is one of the fastest ways to scan a dataset for relationships:

corr = df[["age", "monthly_spend"]].corr()
sns.heatmap(corr, annot=True, cmap="RdBu_r", vmin=-1, vmax=1)
plt.show()

Design Rules That Make Charts Better

• Start bar charts at zero. A truncated y-axis exaggerates differences.
• Avoid pie charts with more than 3 or 4 slices. A sorted bar chart is almost always easier to read.
• Use colour with purpose. Highlight the one bar that matters and grey out the rest.
• One message per chart. Put the insight in the title: "USSD transactions fell 18% after the app relaunch" beats "Transactions by channel".
• Save at high resolution for reports: fig.savefig("chart.png", dpi=200, bbox_inches="tight").

Open in Google Colab

Paste the code blocks above into a new notebook and try changing the chart types, colours and titles. Seaborn and Matplotlib are already available in Colab.

Launch a Colab notebook →