Lesson 10

Working with Figure and Axes Objects

Use plt.subplots to hold the figure and its axes in variables, and learn the set_ methods that replace the plt shortcuts on a real project.

Intermediate25 min

Working with Figure and Axes Objects hero Module 25 Lesson 10

Everything so far has used plt.something, which acts on whichever axes happens to be current. That is comfortable for one chart and fragile for several. The alternative is to keep the objects: ask for the figure and the axes, put them in variables, and say exactly which one you mean. Almost all published Matplotlib code is written this way.

The steps of the object style, from plt.subplots to suptitle

Ask for both objects

plt.subplots() returns a pair: the figure and the axes. Unpack them into two names and you never have to wonder which plot a call will land on.

python

import matplotlib.pyplot as plt

days = ["Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"]
signups = [42, 58, 51, 66, 73, 61, 47]

figure, area = plt.subplots(figsize=(8, 4))
area.plot(days, signups, marker="o")
area.set_title("Signups last week")
area.set_xlabel("day")
area.set_ylabel("people")
area.grid(axis="y", alpha=0.4)

print(type(figure).__name__, "and", type(area).__name__)
plt.show()

Output:

text

Figure and Axes

The names that change

Most of the shortcuts gain a set_ prefix on an axes object. A few keep their names. Learn the pattern once and the whole interface follows.

python

import matplotlib.pyplot as plt

pairs = [
    ("plt.title(...)", "area.set_title(...)"),
    ("plt.xlabel(...)", "area.set_xlabel(...)"),
    ("plt.ylim(...)", "area.set_ylim(...)"),
    ("plt.xticks(...)", "area.set_xticks(...)"),
    ("plt.plot(...)", "area.plot(...)"),
    ("plt.legend(...)", "area.legend(...)"),
    ("plt.suptitle(...)", "figure.suptitle(...)"),
]
for shortcut, method in pairs:
    print(f"{shortcut:<20} {method}")

The translation table:

text

plt.title(...)       area.set_title(...)
plt.xlabel(...)      area.set_xlabel(...)
plt.ylim(...)        area.set_ylim(...)
plt.xticks(...)      area.set_xticks(...)
plt.plot(...)        area.plot(...)
plt.legend(...)      area.legend(...)
plt.suptitle(...)    figure.suptitle(...)

A grid of axes

Pass a number of rows and columns and plt.subplots hands back an array of axes instead of one. Index it like any array, or use .flat to walk every panel in reading order.

python

import matplotlib.pyplot as plt

days = ["Mon", "Tue", "Wed", "Thu", "Fri"]
weekly = {
    "Northgate": [12, 18, 14, 21, 25],
    "Eastport": [9, 11, 15, 13, 12],
    "Westfield": [22, 19, 24, 28, 31],
    "Southbank": [5, 8, 7, 11, 14],
}

figure, grid = plt.subplots(2, 2, figsize=(10, 6))
print("shape of the axes array:", grid.shape)

for area, (depot, values) in zip(grid.flat, weekly.items()):
    area.plot(days, values, marker="o")
    area.set_title(depot)
    area.set_ylim(0, 35)

figure.suptitle("Deliveries by depot", fontsize=15)
figure.tight_layout()
plt.show()

Output:

text

shape of the axes array: (2, 2)

A two by two grid of panels labelled by their array position

Giving every panel the same set_ylim matters here. Without it each panel scales itself, and four charts that look alike would be showing four different ranges.

sharex=True ties the panels to one x range and hides the repeated tick labels, and sharey=True does the same for the value axis. Sharing y is the honest default when panels are meant to be compared.

python

import matplotlib.pyplot as plt

days = ["Mon", "Tue", "Wed", "Thu", "Fri"]
north = [12, 18, 14, 21, 25]
west = [22, 19, 24, 28, 31]

figure, (top, bottom) = plt.subplots(2, 1, figsize=(8, 5), sharex=True, sharey=True)
top.plot(days, north, marker="o")
top.set_title("Northgate")
bottom.plot(days, west, marker="s", color="#EC4899")
bottom.set_title("Westfield")
bottom.set_xlabel("day")

print("same y range:", top.get_ylim() == bottom.get_ylim())
figure.tight_layout()
plt.show()

Output:

text

same y range: True

Key takeaways

plt.subplots() returns the figure and the axes together
On an axes, most plt shortcuts gain a set_ prefix
Rows and columns return an array of axes; .flat walks them in order
sharey keeps panels on one scale so they can honestly be compared
Next: Bar Charts
Previous: Several Plots in One Figure

Loading playground…

Check your understanding