Charts and Visualisation 📈

The simplest chart

import matplotlib      # pip install matplotlib
matplotlib.use("Agg")           # render to a file, no window needed
import matplotlib.pyplot as plt

months = ["Jan", "Feb", "Mar", "Apr", "May", "Jun"]
sales = [100, 120, 118, 135, 150, 162]

fig, ax = plt.subplots(figsize=(8, 4.5))
ax.plot(months, sales, marker="o")
ax.set_title("Monthly sales, 2026")
ax.set_xlabel("Month")
ax.set_ylabel("Sales (£000s)")
ax.grid(alpha=0.3)

fig.tight_layout()
fig.savefig("sales.png", dpi=150)
print("written to sales.png")

The pattern is always the same: make a figure and axes, draw on the axes, label everything, save or show. matplotlib.use("Agg") tells it to render to a file rather than open a window, which is what you want on a server or in a script.

Choosing the right chart

Several charts at once

import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
import numpy as np

rng = np.random.default_rng(seed=42)
data = rng.normal(loc=100, scale=15, size=500)

fig, axes = plt.subplots(1, 3, figsize=(14, 4))

axes[0].hist(data, bins=30, edgecolor="white")
axes[0].set_title("Distribution")

axes[1].boxplot(data, vert=False)
axes[1].set_title("Spread and outliers")

axes[2].scatter(data[:-1], data[1:], alpha=0.4, s=12)
axes[2].set_title("Each value against the next")

for ax in axes:
    ax.grid(alpha=0.3)

fig.suptitle("Three views of the same 500 numbers")
fig.tight_layout()
fig.savefig("three-views.png", dpi=150)
print("saved")

Note the seeded random generator: default_rng(seed=42) makes the figure reproducible, which matters as much for a chart in a report as it does for a test.

The chart nobody labels properly

import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt

regions = ["East", "North", "South", "West"]
totals = [4200, 2090, 1490, 980]

fig, ax = plt.subplots(figsize=(8, 4.5))
bars = ax.barh(regions, totals, color="#4584b6")

ax.set_title("Total sales by region, Q2 2026")
ax.set_xlabel("Sales (£)")
ax.bar_label(bars, fmt="£{:,.0f}", padding=4)
ax.set_xlim(0, max(totals) * 1.15)
ax.spines[["top", "right"]].set_visible(False)

fig.tight_layout()
fig.savefig("regions.png", dpi=150)
print("saved")

A chart is finished when it has:

• a title that states the finding, not just the subject;
• axis labels with units;
• a source and a date, if anyone else will see it;
• no more ink than the data needs. Remove gridlines, borders and legends that earn nothing.

How charts lie

import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt

months = ["Jan", "Feb", "Mar", "Apr"]
values = [100, 102, 101, 104]

fig, (left, right) = plt.subplots(1, 2, figsize=(11, 4))

left.bar(months, values, color="#c92a2a")
left.set_ylim(99, 105)
left.set_title("'Explosive growth!'")

right.bar(months, values, color="#4584b6")
right.set_ylim(0, 120)
right.set_title("The same data, honest axis")

fig.tight_layout()
fig.savefig("truncated.png", dpi=150)
print("saved")

The other reliable ways to mislead, in case you meet them:

• Cherry-picked ranges. Any trend can be reversed by choosing the start date.
• Two y-axes. Almost always used to imply a relationship between unrelated series, and the scales can be tuned until the lines agree.
• Area for a single number. Doubling a circle's radius quadruples its area, so the reader sees a 4x change in a 2x quantity.
• Correlation presented as cause. Ice cream sales and drownings rise together. Neither causes the other; summer causes both.

Other tools

import pandas as pd

df = pd.DataFrame({"month": ["Jan", "Feb", "Mar"], "sales": [100, 120, 118]})

# the fastest possible look at some data
ax = df.plot(x="month", y="sales", kind="bar", title="Sales", figsize=(6, 3))
ax.figure.savefig("quick.png")

import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
import random

random.seed(42)
totals = [random.randint(1, 6) + random.randint(1, 6) for _ in range(1000)]

fig, ax = plt.subplots(figsize=(8, 4.5))
ax.hist(totals, bins=range(2, 14), align="left", rwidth=0.85, color="#4584b6")

ax.set_title("Two dice, 1,000 rolls: seven is the most likely total")
ax.set_xlabel("Total of both dice")
ax.set_ylabel("Number of rolls")
ax.set_xticks(range(2, 13))
ax.spines[["top", "right"]].set_visible(False)

fig.tight_layout()
fig.savefig("dice.png", dpi=150)
print("saved dice.png")