TopolearnEvery customization you will ever do is a method on a Figure or an Axes, so this mental model makes the rest of the library predictable.
A Matplotlib chart is a small tree of objects. At the top is the Figure: the whole canvas, with a size in inches and a resolution in dots per inch. Inside it live one or more Axes: a plotting area with an x axis, a y axis, a title and the data you draw. Note the spelling: an Axes is one plot; an Axis is one of its two number lines.
The usual entry point is matplotlib.pyplot, imported as plt. The idiomatic start is:
fig, ax = plt.subplots()which creates a Figure with a single Axes and returns both. You then call methods on ax to draw (ax.plot, ax.bar) and to decorate (ax.set_title, ax.set_xlabel). Finally you either show the figure in a window or notebook with plt.show(), or write it to a file with fig.savefig("chart.png").
figsize=(width, height)is in inches;dpiturns inches into pixels when saving a raster image.- Each Axes has its own data limits, so two side-by-side plots can use different scales.
- pyplot keeps a notion of the "current" figure and axes, which is what the short
plt.plot(...)style draws on.
When nothing appears, it is usually because the script never called plt.show(), or it is running without a display (a server), where you should save to a file instead.
Example
import matplotlib
matplotlib.use("Agg") # no window needed, e.g. on a server
import matplotlib.pyplot as plt
fig, ax = plt.subplots(figsize=(6, 4))
ax.plot([1, 2, 3, 4], [1, 4, 9, 16])
ax.set_title("Squares")
print(type(fig).__name__) # Figure
print(type(ax).__name__) # Axes
print(fig.get_size_inches()) # [6. 4.]
print(len(fig.axes)) # 1
fig.savefig("squares.png", dpi=100) # 600 x 400 pixelsRemember
- A Figure is the canvas; an Axes is one plot inside it
fig, ax = plt.subplots()is the standard way to start- Draw and decorate by calling methods on
ax; save withfig.savefig
Watch out for
- Confusing Axes (a plot) with Axis (the x or y number line).