← Matplotlib·Lesson 1·10 min·0/3 exercises

Figure, Axes and Artist

Every matplotlib question is really “which object owns this?”

FigureAxesArtistfig.suptitleax.spinesax.xaxis

Watch it happen

Play it through, or step back and forth yourself.

fig
the whole canvas
ax
one plotting area
ax.plot(...)
returns the Line2D artist
ax.set_title(...)
text belongs to the Axes
ax.spines / ax.xaxis
the frame and the scales

A finished matplotlib chart. It looks like one thing and it's actually a small tree of objects — and knowing which object owns what is most of learning matplotlib.

The idea

matplotlib has a reputation for being fiddly, and most of that comes from one thing: people learn the commands before they learn the objects. Learn the objects first and the commands stop needing to be memorised.

There are three, and they nest.

Figure — the canvas

The outermost container. It has a size in inches and a resolution in dots per inch; multiply them and you have the pixel dimensions of the file you save. It draws nothing itself.

fig = plt.figure(figsize=(6, 3.5), dpi=150)   # 900 x 525 pixels
fig.suptitle("A title for the whole figure")

Axes — one plotting area

Inside the Figure sit one or more Axes. An Axes is a single plot: a rectangle with its own x scale, y scale, title and labels. This is where data goes, and it's the object you'll interact with most.

The name is the single worst thing about matplotlib's API. An Axes is a whole subplot. An Axis (ax.xaxis) is one of its two scales. They differ by one letter and mean completely different things.

fig, ax = plt.subplots()          # one Figure, one Axes
fig, axes = plt.subplots(2, 2)    # one Figure, four Axes

Artist — everything drawn

Every visible thing is an Artist: the line, the markers, the bars, the text, the legend, the tick marks, the frame. ax.plot() doesn't so much draw as create an artist and hand it back:

line, = ax.plot(days, delhi)   # a Line2D
line.set_color("#66aaf9")       # still editable afterwards
line.set_linewidth(3)

That trailing comma is worth noticing — plot returns a list, because one call can draw several lines.

Who owns what

Almost everything you'll want to change belongs to the Axes:

ax.set_title(...)      ax.set_xlabel(...)     ax.legend()
ax.set_xlim(...)       ax.set_xticks(...)     ax.grid(...)
ax.spines["top"]       ax.xaxis               ax.yaxis

The Figure owns only the things that are about the whole canvas — fig.suptitle(), fig.savefig(), fig.set_size_inches(), fig.tight_layout().

So when you don't know how to change something, ask which object owns it. Tick colour? The axis, via the Axes. Frame? A spine, via the Axes. Overall size? The Figure. That one question replaces a lot of searching.

Looking inside

fig.axes            # every Axes in this figure
ax.get_figure()     # back up to the Figure
ax.lines            # the Line2D artists on this Axes
ax.get_children()   # everything the Axes owns

ax.get_children() is genuinely useful when you're stuck: it lists every artist on the Axes, and the thing you want to change is in there somewhere.

See it run

The lesson's code, ready to run and to fiddle with.

Putting the kettle on…

Starting up…

Worked example

not graded

Already written and ready to go — press Run to see what it does, then change a number, a column name, anything, and run it again.

tryprinting ax.get_children() to see everything the Axes owns.

Press Run — the output appears here.

Your turn

3 exercises. Write the code yourself, then press Check — a nudge and the answer are there if you want them.

Exercise 1

Create a figure and axes with plt.subplots(), and return the type name of the first object — it should be Figure.

your answer

Exercise 2

Make a figure with four Axes in a 2×2 grid, and return how many Axes the figure contains.

your answer

Exercise 3

Plot delhi against days and return the type name of the artist that ax.plot() hands back. Remember it returns a list.

your answer