Real Python Matplotlib Tutorial
Resources¶
import matplotlib.pyplot as plt
import numpy as np
np.random.seed(444)
-
One important big-picture matplotlib concept is its object hierarchy.
-
A “hierarchy” here means that there is a tree-like structure of matplotlib objects underlying each plot.
-
A Figure object is the outermost container for a matplotlib graphic, which can contain multiple Axes objects. One source of confusion is the name: an Axes actually translates into what we think of as an individual plot or graph (rather than the plural of “axis,” as we might expect).
fig, _ = plt.subplots()
type(one_tick := fig.axes[0].yaxis.get_major_ticks()[0])
-
Notice that we didn’t pass arguments to subplots() here. The default call is subplots(nrows=1, ncols=1)
fig, ax = plt.subplots()
type(ax)
-
We can call its instance methods to manipulate the plot similarly to how we call pyplots functions. Let’s illustrate with a stacked area graph of three time series
(rng := np.arange(50))
(rnd := np.random.randint(0, 10, size=(3, rng.size)))
(yrs := 1950 + rng)
fig, ax = plt.subplots(figsize=(5, 3))
ax.stackplot(yrs, rng + rnd, labels=["Eastasia", "Eurasia", "Oceania"])
ax.set_title("Combined debt growth over time")
ax.legend(loc="upper left")
ax.set_ylabel("Total debt")
ax.set_xlim(xmin=yrs[0], xmax=yrs[-1])
fig.tight_layout()
fig
Let’s look at an example with multiple subplots (Axes) within one Figure, plotting two correlated arrays that are drawn from the discrete uniform distribution:
(x := np.random.randint(low=1, high=11, size=50))
(y := x + np.random.randint(1, 5, size=x.size))
(data := np.column_stack((x, y)))
fig, (ax1, ax2) = plt.subplots(nrows=1, ncols=2, figsize=(8, 4))
ax1.scatter(x=x, y=y, marker="o", c="r", edgecolor="b")
ax1.set_title("Scatter: $x$ versus $y$")
ax1.set_xlabel("$x$")
ax1.set_ylabel("$y$")
ax2.hist(data, bins=np.arange(data.min(), data.max()), label=("x", "y"))
ax2.legend(loc=(0.65, 0.8))
ax2.set_title("Frequencies of $x$ and $y$")
ax2.yaxis.tick_right()
-
Text inside dollar signs utilizes TeXmarkup to put variables in italics.
-
Because we’re creating a “1x2” Figure, the returned result of plt.subplots(1, 2) is now a Figure object and a NumPy array of Axes objects. (You can inspect this with fig, axs = plt.subplots(1, 2) and taking a look at axs.)
fig
tuple(fig.axes[i] is ax for i, ax in zip(range(2), (ax1, ax2)))
-
Taking this one step further, we could alternatively create a figure that holds a 2x2 grid of Axes objects
fig, ax = plt.subplots(nrows=2, ncols=2, figsize=(7, 7))
-
Now, what is ax? It’s no longer a single Axes, but a two-dimensional NumPy array of them
type(ax)
ax
ax1, ax2, ax3, ax4 = ax.flatten()
To illustrate some more advanced subplot features, let’s pull some macroeconomic California housing data extracted from a compressed tar archive, using io, tarfile, and urllib from Python’s Standard Library.
import tarfile
from urllib.request import urlretrieve
from pathlib import Path
filepath, response = urlretrieve(
"http://www.dcc.fc.up.pt/~ltorgo/Regression/cal_housing.tgz"
)
filepath
with tarfile.open(name=filepath, mode="r") as archive:
housing = np.loadtxt(
archive.extractfile("CaliforniaHousing/cal_housing.data"), delimiter=","
)
housing
(y := housing[:, -1])
The property T is an accessor to the method transpose().
pop, age = housing[:, [4, 7]].T
pop, age
Next let’s define a “helper function” that places a text box inside of a plot and acts as an “in-plot title”:
def add_titlebox(ax, text):
ax.text(
0.55,
0.8,
text,
horizontalalignment="center",
transform=ax.transAxes,
bbox=dict(facecolor="white", alpha=0.6),
fontsize=12.5,
)
return ax
gridsize = (3, 2)
fig = plt.figure(figsize=(12, 8))
ax1 = plt.subplot2grid(gridsize, (0, 0), colspan=2, rowspan=2)
ax2 = plt.subplot2grid(gridsize, (2, 0))
ax3 = plt.subplot2grid(gridsize, (2, 1))
ax1.set_title("Home value as a function of home age & area population", fontsize=14)
sctr = ax1.scatter(x=age, y=pop, c=y, cmap="RdYlGn")
plt.colorbar(sctr, ax=ax1, format="$%d")
ax1.set_yscale("log")
ax2.hist(age, bins="auto")
ax3.hist(pop, bins="auto", log=True)
add_titlebox(ax2, "Histogram: home age")
add_titlebox(ax3, "Histogram: area population (log scl.)")
fig
[plt.figure(i) for i in plt.get_fignums()]
plt.get_fignums()
While ax.plot() is one of the most common plotting methods on an Axes, there are a whole host of others, as well. (We used ax.stackplot() above. You can find the complete list here.)
Methods that get heavy use are imshow() and matshow(), with the latter being a wrapper around the former. These are useful anytime that a raw numerical array can be visualized as a colored grid.
(x := np.diag(np.arange(2, 12))[::-1])
x[np.diag_indices_from(x[::-1])] = np.arange(2, 12)
x
(x2 := np.arange(x.size).reshape(x.shape))
sides = (
"left",
"right",
"top",
"bottom",
)
nolabels = {s: False for s in sides}
nolabels.update({f"label{s}": False for s in sides})
nolabels
from mpl_toolkits.axes_grid1.axes_divider import make_axes_locatable
with plt.rc_context(rc={"axes.grid": False}):
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(8, 4))
ax1.matshow(x)
img2 = ax2.matshow(x2, cmap="RdYlGn_r")
for ax in (ax1, ax2):
ax.tick_params(axis="both", which="both", **nolabels)
for i, j in zip(*x.nonzero()):
ax1.text(j, i, x[i, j], color="white", ha="center", va="center")
divider = make_axes_locatable(ax2)
cax = divider.append_axes("right", size="5%", pad=0)
plt.colorbar(img2, cax=cax, ax=[ax1, ax2])
fig.suptitle("Heatmaps with `Axes.matshow`", fontsize=16)
The pandas library has become popular for not just for enabling powerful data analysis, but also for its handy pre-canned plotting methods. Interestingly though, pandas plotting methods are really just convenient wrappers around existing matplotlib calls.
import pandas as pd
(s := pd.Series(np.arange(5), index=list("abcde")))
(ax := s.plot())
type(ax)
id(plt.gca()), id(ax)
import matplotlib.transforms as mtransforms
url = "https://fred.stlouisfed.org/graph/fredgraph.csv?id=VIXCLS"
vix = pd.read_csv(
url, index_col=0, parse_dates=True, infer_datetime_format=True, squeeze=True
).dropna()
vix = vix[vix != "."]
(ma := vix.rolling("90d").mean())
{type(item) for item in vix}
vix_ = pd.Series([float(item) for item in vix])
state = pd.cut(ma, bins=[-np.inf, 14, 18, 24, np.inf], labels=range(4))
cmap = plt.get_cmap("RdYlGn_r")
ma.plot(color="black", linewidth=1.5, marker="", figsize=(8, 4), label="VIX 90d MA")
ax = plt.gca() # Get the current Axes that ma.plot() references
ax.set_ylabel("90d moving average: CBOE VIX")
ax.set_title("Volatility Regime State")
ax.grid(False)
ax.legend(loc="upper center")
ax.set_xlim(xmin=ma.index[0], xmax=ma.index[-1])
trans = mtransforms.blended_transform_factory(ax.transData, ax.transAxes)
for i, color in enumerate(cmap([0.2, 0.4, 0.6, 0.8])):
ax.fill_between(ma.index, 0, 1, where=state == i, facecolor=color, transform=trans)
ax.axhline(
vix_.mean(),
linestyle="dashed",
color="xkcd:dark grey",
alpha=0.6,
label="Full-period mean",
marker="",
)
