Charts with pandas and matplotlib

01The problem

Bisi's analysis is right, but the managing director won't read a table of 18 monthly totals. He will look at one good chart for five seconds. In those five seconds he should see the answer, not work it out.

A chart earns its place when it makes one point obvious. This lesson covers the two charts that do most of an analyst's work, the line chart (how something changed over time) and the sorted bar chart (how things compare), and the small details that make them readable: the right scale, labels in millions, and a title that says what the chart shows.

02The concept

pandas draws with matplotlib

Every Series and DataFrame has a .plot() method that uses matplotlib, Python's main charting library, underneath. You import its plotting part as plt to adjust the chart:

import matplotlib.pyplot as plt

Pick the chart for the question

Horizontal bars (barh) are easier to read when the labels are long, like region or product names. Sort them so the eye goes straight from biggest to smallest.

Avoid pie charts for more than three or four slices, and avoid 3D charts entirely. People can't compare angles or perspective accurately; they can compare bar lengths.

Make it readable

• ax = series.plot(...) returns the chart's axes. You use it to set the title and labels: ax.set_title(...), ax.set_xlabel(...), ax.set_ylabel(...).
• Large naira values: divide by 1,000,000 before plotting and label the axis "₦ millions". 1e6 is Python's shorthand for 1,000,000.
• Bar charts start at zero. A bar axis that starts at ₦15m makes a small difference look huge.
• figsize=(8, 4) sets the size in inches; wide and short suits time series.

Titles that state the finding

"Revenue by region" describes the chart. "North West revenue fell 47% while South West grew 79%" tells the reader what to take away. That second kind, an action title, is the most useful habit in this lesson.

Saving

plt.savefig("revenue_trend.png", dpi=200, bbox_inches="tight") saves the current chart. Call it before plt.show(). bbox_inches="tight" stops the labels being cut off.

03Example

A monthly trend with its 3-month rolling average:

import pandas as pd
import matplotlib.pyplot as plt

base = "https://academy.cloudtechanalytics.com/datasets/sales/"
orders = pd.read_csv(base + "orders.csv", parse_dates=["order_date"])
customers = pd.read_csv(base + "customers.csv")
orders["revenue"] = orders["quantity"] * orders["unit_price"] * (1 - orders["discount_pct"] / 100)

monthly = orders.groupby(orders["order_date"].dt.to_period("M"))["revenue"].sum() / 1e6

ax = monthly.plot(figsize=(9, 4), marker="o", label="Monthly revenue")
monthly.rolling(3).mean().plot(ax=ax, linewidth=3, label="3-month average")
ax.set_title("Revenue is growing: H1 2026 is 19% above H1 2025, with a December peak")
ax.set_xlabel("")
ax.set_ylabel("₦ millions")
ax.legend()
plt.savefig("revenue_trend.png", dpi=200, bbox_inches="tight")
plt.show()

Passing ax=ax draws the second line on the same chart.

And a sorted bar chart of H1 growth by region:

full = orders.merge(customers, on="customer_id", how="left", validate="many_to_one")
full["year"] = full["order_date"].dt.year
h1 = full[full["order_date"].dt.month <= 6]
by_region = h1.pivot_table(index="region", columns="year", values="revenue", aggfunc="sum")
growth_pct = ((by_region[2026] / by_region[2025] - 1) * 100).sort_values()

colours = ["#b4442c" if g < 0 else "#4d6b57" for g in growth_pct]
ax = growth_pct.plot.barh(figsize=(8, 4), color=colours)
ax.axvline(0, color="black", linewidth=0.8)
ax.set_title("North West revenue fell 47% in H1 2026; South West grew 79%")
ax.set_xlabel("Change from H1 2025 (%)")
ax.set_ylabel("")
plt.show()

Colouring the falls differently makes the point before anyone reads a number.

04Walkthrough

1. Run the trend chart. Then remove / 1e6 and run it again: the axis fills with numbers like 6e7. Put it back.
2. Change the title to "Monthly revenue" and compare. Which one would the managing director remember?
3. Run the region bar chart. Try .plot.bar() instead of .plot.barh() and see how the region names squeeze together. Horizontal wins.
4. Draw a histogram of order-line revenue: orders["revenue"].plot.hist(bins=30). Most lines are small, with a long tail of big ones: the "skew" you met with credit limits in lesson 3.
5. Category mix by year as a stacked bar:

products = pd.read_csv(base + "products.csv")
with_category = orders.merge(products, on="product_id", how="left", validate="many_to_one")
with_category["year"] = with_category["order_date"].dt.year
mix = with_category.pivot_table(index="year", columns="category", values="revenue", aggfunc="sum") / 1e6
ax = mix.plot.bar(stacked=True, figsize=(6, 4))
ax.set_title("Household and Personal care lead in both years")
ax.set_ylabel("₦ millions")
plt.show()

6. Note that 2026 is a shorter bar because it's half a year. If you show this chart, say so in the title or a note, or chart H1 against H1 instead.
7. Save one chart with plt.savefig and download it from Colab's Files panel.

05Practice

06Challenge

07Check your understanding