Build Bar & Pie Charts

You’ve cleaned your data, grouped it, and summarized it — but when you present your findings to stakeholders, all you get back is a blank stare. Raw numbers in a table rarely tell a story. That’s the pain this lesson solves: turning tidy data into bar charts and pie charts that make trends and proportions instantly obvious. By the end, you’ll be able to build bar charts and pie charts in Python with Matplotlib, tweak them for clarity, and choose the right chart for the story your data is telling — no design degree required.

The problem this lesson solves

Spreadsheets and DataFrames are great for computation, but humans are visual creatures. A table of sales by region might contain a clear insight — “the West region dominates” — but your audience won’t see it until you draw a picture. This lesson tackles the exact pain of communicating data insights through two of the most common chart types: bar charts for comparing categories and pie charts for showing parts of a whole.

Without these skills, you’ll find yourself copying numbers into a slides deck or explaining trends in prose — slow, error-prone, and unconvincing. Worse, you might try to eyeball a chart from a screenshot or use a spreadsheet tool, breaking your Python workflow. This lesson gives you a direct path from pandas DataFrame to publication-ready chart in a few lines of code.

Core concept / mental model

Think of a bar chart as a comparing machine: each bar’s height encodes a value, so your eye can instantly judge which category is bigger. A pie chart is a sharing machine: each slice’s area encodes a proportion, so you see how the whole splits into parts.

In Matplotlib, both charts are built from the same canvas: a Figure holding one or more Axes. You call a single method — bar() or pie() — and then layer on labels, titles, and styling. The mental model is simple: data → Figure → Axes → plot method → labels → show/save.

How it works step by step

1. Set up the data

Every chart starts with clean, structured data. For a bar chart, you need categories and numeric values. For a pie chart, you need slice labels and numeric sizes. This often comes from a DataFrame using groupby() or value_counts().

2. Create the figure and axes

Matplotlib’s plt.subplots() gives you a canvas and a plotting area. Always use it — it’s the modern, explicit way to control your chart.

3. Plot the chart

Choose the right method: - ax.bar(x, height) for vertical bars - ax.barh(x, width) for horizontal bars - ax.pie(sizes, labels=labels) for a pie chart

4. Add context

Titles, axis labels, and legends turn a raw chart into a message. Never skip labels — a chart without them is noise.

5. Show or save

Use plt.show() in a notebook or fig.savefig('chart.png') for reports.

The cause → effect logic: better labels → better comprehension; correct chart choice → clearer insight; consistent styling → professional presentation.

Hands-on walkthrough

Let’s build both charts from a real dataset. We’ll use a sample of sales by region and product category.

Setup

import matplotlib.pyplot as plt
import pandas as pd

# Sample data: sales by region
sales = pd.DataFrame({
    'region': ['North', 'South', 'East', 'West'],
    'sales': [5200, 4100, 3800, 6900]
})
print(sales)

Output:

  region  sales
0  North   5200
1  South   4100
2  East    3800
3  West    6900

Build a bar chart

fig, ax = plt.subplots(figsize=(8, 5))
ax.bar(sales['region'], sales['sales'], color=['#4C72B0', '#55A868', '#C44E52', '#8172B2'])
ax.set_title('Sales by Region', fontsize=16)
ax.set_xlabel('Region')
ax.set_ylabel('Sales (USD)')
ax.grid(axis='y', linestyle='--', alpha=0.7)
plt.tight_layout()
plt.show()

You should see a clean bar chart with four colored bars, a title, and labeled axes. The West region clearly stands out.

Build a pie chart

fig, ax = plt.subplots(figsize=(7, 7))
sizes = sales['sales']
labels = sales['region']
ax.pie(sizes, labels=labels, autopct='%1.1f%%', startangle=90, colors=['#4C72B0', '#55A868', '#C44E52', '#8172B2'])
ax.set_title('Sales Distribution by Region', fontsize=16)
plt.show()

Output: a pie chart with percentage labels on each slice, starting from the top (90°) and going clockwise.

Add data labels to bar charts

fig, ax = plt.subplots(figsize=(8, 5))
bars = ax.bar(sales['region'], sales['sales'])
ax.bar_label(bars, fmt='${:,.0f}')
ax.set_title('Sales by Region with Labels')
plt.show()

ax.bar_label() adds value labels on top of each bar — a pro move that saves your audience from reading the y-axis.

Compare options / when to choose what

Pro tip: When in doubt, pick a bar chart. Research shows humans compare bar lengths more accurately than pie slice angles. Use a pie only when the whole-to-part story is the headline.

Variations to consider

• Horizontal bar chart (ax.barh()) for long category names
• Grouped bar chart for comparing multiple series side-by-side
• Donut chart (a pie with a hole) for a modern look and center labels

Troubleshooting & edge cases

Pie chart percentages don’t sum to 100%

This happens when your values aren’t the whole. Ensure you pass all parts, or the pie will mislead. Use autopct='%1.1f%%' — Matplotlib computes percentages internally, but only from the values you give it.

Overlapping labels

Long labels on a pie can collide. Fix by: - Using plt.tight_layout() - Increasing figure size - Rotating labels with rotatelabels=True - Switching to a bar chart if labels are too long

Bars look too thin or too thick

Adjust the width parameter in ax.bar() (default 0.8). A value like 0.6 makes bars slimmer and reduces crowding.

Negative values in bar chart

Bar charts handle negatives fine, but the baseline moves. Use ax.axhline(0, color='black', linewidth=0.8) to draw a zero line for clarity.

Pie chart with zero or negative values

Never use a pie with zero or negative values — slices vanish or mislead. Clean your data first, or switch to a bar chart.

Chinese/Unicode labels show as boxes

Set a font that supports your characters:

plt.rcParams['font.family'] = 'sans-serif'
plt.rcParams['font.sans-serif'] = ['SimHei', 'DejaVu Sans']

Colormap too similar

Use a distinct colormap or pass explicit colors as a list. Avoid default colors for adjacent slices that are hard to tell apart.

What you learned & what's next

You now know how to build bar charts and pie charts in Python: you’ve set up Matplotlib, plotted both chart types, added labels and titles, and styled them for clarity. You can choose between a bar and a pie based on your data story, and you’ve picked up troubleshooting skills for common issues like overlapping labels and misleading pies.

These charts are your first step into data visualization. Next up, you’ll learn how to create histograms and box plots to explore the distribution of your data — a natural follow-on that reveals spread, outliers, and shape. You’ll reuse everything you learned here: the same subplots() canvas, the same ax.set_title() pattern, and the same instinct to question whether your chart is telling the truth.

Open a notebook and practice: take a DataFrame you already have, group it by a categorical column, and build both a bar and a pie chart. Then ask yourself — which one tells the story better? That judgment, not the code, is the skill that will make you a better data scientist.