Lunch Time Python¶

Lunch 7: matplotlib¶

matplotlib is a plotting library for Python and the NumPy library. It is easy to use and can be used to generate scatter or bar plots, density maps and even 3D plots in publication quality.

Press Spacebar to go to the next slide (or ? to see all navigation shortcuts)

Lunch Time Python, Scientific Software Center, Heidelberg University

Advantages of matplotlib:¶

  • matplotlib is mostly used in conjunction with pyplot - a matplotlib module - that provides an easy-to-use interface
  • Very versatile, works with many types of data
  • Directly plot NumPy functions and arrays
  • Many export options
  • Customizable
  • Can be used with additional toolkits that extend the functionality, like seaborn

matplotlib installation¶

Available via pip:
python -m pip install -U matplotlib
Or install via conda:
conda install matplotlib

Basic plots¶

Plot the sin and cos over a range of angles. For this, we also need numpy.

In [1]:
import numpy as np
import matplotlib.pyplot as plt
In [2]:
xvals = np.arange(0, 2 * np.pi, 0.1)
In [3]:
plt.plot(xvals, np.sin(xvals))
plt.plot(xvals, np.cos(xvals))
Out[3]:
[<matplotlib.lines.Line2D at 0x7f1f60037ca0>]

The default settions already look quite nice!

Running inside a script¶

Or more generally, if you want to suppress the output (return) of the plt() function.

In [4]:
plt.plot(xvals, np.sin(xvals))
plt.plot(xvals, np.cos(xvals))
plt.show()

plt.show() closes the plot; if you plot using a script and not a notebook, place one plt.show() command at the end of your script.

Running inside a notebook: static images¶

You can plot static images inside your notebook using the %matplotlib inline magic: You only need to run this once. It is not always necessary to put this, but it makes it clear which Matplotlib backend should be used.

In [5]:
%matplotlib inline
In [6]:
plt.plot(xvals, np.sin(xvals))
plt.plot(xvals, np.cos(xvals))
plt.show()

Scatter plot¶

In [7]:
xvals = np.random.randint(10, size=10)
yvals = np.random.randint(10, size=10)
plt.scatter(xvals, yvals)
plt.show()

Bar plot¶

In [8]:
xvals = np.linspace(1, 10, 10)
yvals = np.random.randint(10, size=10)
plt.bar(xvals, yvals)
plt.show()

Customizing plots¶

In [9]:
plt.bar(xvals, yvals, label="Random series")
plt.legend()
plt.show()

Customizing plots¶

In [10]:
plt.bar(xvals, yvals, label="Random series")
plt.legend(fontsize=16, loc="upper right")
plt.xlabel("Integer", fontsize=18)
plt.ylabel("Magnitude", fontsize=22, color="red")
plt.title("My custom plot", fontsize=22)
plt.show()

Customizing plots¶

In [11]:
xvals = np.arange(0, 2 * np.pi, 0.1)
plt.plot(xvals, np.sin(xvals), marker="x", markevery=10, color="blue")
plt.plot(xvals, np.cos(xvals), marker="<", color="black", alpha=0.5)
plt.show()

You can set x- and y-limits, annotate points on the plot, change the axis position and intersection, add a grid in the background, ...
A list of available markers is found here and named colors here.

Figure class: Advanced plots with the artist layer¶

A Figure in matplotlib is the whole plot (or window in the user interface) and can contain multiple plots. By accessing the Artist layer ("object-based plotting"), you can access more customizing options than with the basic plt.xxx Scripting layer ("procedural plotting") (see https://matplotlib.org/1.5.1/faq/usage_faq.html#parts-of-a-figure).

This also allows you to include multiple plots in one Figure.

In [12]:
fig = plt.figure(figsize=(12, 10))  # width = 12 inches and height = 10 inches
# create one axes object
ax1 = fig.add_subplot(211)  # (2, 1, 1) no of rows, no of columns, no of plots
ax1.plot(xvals, np.sin(xvals), marker="x", color="blue")
plt.show()

It is also possible to use add_axes() instead of add_subplot(), but not recommended as with the latter, matplotlib takes care of the exact position of the axes in the figure.

Several subplots¶

Using add_subplot(), we can add one axes object at a time:

In [13]:
fig = plt.figure(figsize=(12, 10))
ax1 = fig.add_subplot(221)
ax1.plot(xvals, np.sin(xvals), marker="x", color="blue")
ax2 = fig.add_subplot(222)
ax2.plot(xvals, np.cos(xvals), marker="x", color="blue")
ax3 = fig.add_subplot(223)
ax3.plot(xvals, np.tan(xvals), marker="x", color="blue")
ax4 = fig.add_subplot(224)
ax4.plot(xvals, np.tanh(xvals), marker="x", color="blue")
plt.show()

A more practical way to add multiple subplots¶

In [14]:
fig, ax = plt.subplots(figsize=(12, 10), nrows=2, ncols=2)
ax[0, 0].plot(xvals, np.sin(xvals), marker="x", color="blue")
ax[0, 1].plot(xvals, np.cos(xvals), marker="x", color="blue")
ax[1, 0].plot(xvals, np.tan(xvals), marker="x", color="blue")
ax[1, 1].plot(xvals, np.tanh(xvals), marker="x", color="blue")
plt.savefig("my_figure.pdf", bbox_inches="tight")
plt.savefig("my_figure.jpg", dpi=300, bbox_inches="tight")
plt.show()