Ctrl+K

Part II: Matplotlib

Part II: Matplotlib#

3. Matplotlib#

Matplotlib is the most popular plotting library for Python.

We will take a look at several examples below, but the Matplotlib gallery is a great place to search for snippets of code to make your plots look professional.

3.1 Importing Matplotlib#

To use Matplotlib, we first need to import it. Matplotlib has many sub-packages, so we will import the pyplot sub-package.

# Import matplotlib
from matplotlib import pyplot as plt

# Import numpy again too
import numpy as np

Now let’s plot some graphs. To plot a graph, you will need x coordinates and y coordinates.

Let’s plot a sine wave. We can use NumPy to do all sorts of mathematical operations, like computing the sine of a variable.

# Create our x coordinates (note the np.sine() function takes radians as an input)
x = np.linspace(-2*np.pi, 2*np.pi, 200)

# Our y will simply be sin(x)
y = np.sin(x)

We now have our x and y coordinates - we can plot the graph. What do you think it will look like?

# Make a simple line plot
plt.plot(x, y)
Hide code cell output 
[<matplotlib.lines.Line2D at 0x1135a30a0>]
../../_images/e824f079e434c76c4b8bdd80692218726e7bb8d872799003feda35681501022c.png

There is so much styling of plots that you can do and we will by no means cover everything, but we will take a look at a few of the available options.

First, we can change the line to a series of points using markers:

# Make a simple plot using markers ('o')
plt.plot(x, y, 'o')
Hide code cell output 
[<matplotlib.lines.Line2D at 0x11374aec0>]
../../_images/e4e49e062d99a7111410b44f6bf0aabb205e6b8b7ef022a70a332978c824a756.png

The default plotting colour is blue, but you can change this. Below I have changed the colour - can you guess what colour will appear based on the code?

# Make a simple plot using green markers
plt.plot(x, y, 'go')
Hide code cell output 
[<matplotlib.lines.Line2D at 0x1137c7b50>]
../../_images/adfdc7af59e012dc7a576d0c7246f5a099195a5b4c6a75792c35f53ed6b1afd5.png

What will the next example look like?

# Make a simple plot using red markers
plt.plot(x, y, 'r+')
Hide code cell output 
[<matplotlib.lines.Line2D at 0x11382fd60>]
../../_images/608082b75571b0be6f0ea49ecfa45898ed577e248f312eaa95255dc911af83d6.png

The syntax used above to style our plots uses a format string:

Format string

Format string is with the format of fmt = ‘[color][marker][line]’ For example, r+ simply means red with + marker. Full documentation can be found at this link under Notes secion. You can plot multiple sets of data on the same graph.

# Plot with two curves
plt.plot(x, y, 'r+')
plt.plot(x, np.cos(x), 'go', linewidth=4)
Hide code cell output 
[<matplotlib.lines.Line2D at 0x1138cca00>]
../../_images/1bcc3fa0415d9948c8c26d2347abb1fc54327116e9ac4957214a69445f00a568.png

If we have two or more curves in the same plot, as above, how can we convey to the reader which line is which? We can use legends. To use a legend in Matplotlib, you indicate the legend label in the plotting command and then invoke the legend function separately include the legend in the plot.

# Plot with two curves and a legend
plt.plot(x, y, 'r+', label="sine")
plt.plot(x, np.cos(x), 'go', linewidth=4, label="cosine")
plt.legend(loc="lower left")

<matplotlib.legend.Legend at 0x1139290f0>
../../_images/3be9cc4d28feff41f2c3a46e92b2054e877984ae77d7c95320b4608ba3881a8c.png

Finally, no plot is complete without a title and labels on the axes. Let’s take a look at how to add these.

# Plot with two curves, a legend and labels
plt.plot(x, y, 'r+', label="sine")
plt.plot(x, np.cos(x), 'go', linewidth=4, label="cosine")
plt.legend(loc="lower left")

# add labels
plt.title("Sine and Cosine Waves")
plt.ylabel("Amplitude")
plt.xlabel("Time")

Text(0.5, 0, 'Time')
../../_images/f5937b9dde3d6ef48fb336286c25802eb59f4292a75aed7d0a2cdb7964724280.png

1.3 Plotting 2-Dimensions#

Since NumPy is all about multi-dimensional data, we will now take a look at how to create 2-D plots. We will first make some plots using simple arrays and then we’ll take a look at some real data in the next section.

# create three arrays, a, b and mesh
a = np.ones((1,1000))
b = np.linspace(-20, 20, 1000)
print(a.shape, b.shape)
mesh = a * b
print(mesh.shape)

(1, 1000) (1000,)
(1, 1000)

# plot mesh in in 2-D
plt.pcolormesh(mesh)
plt.colorbar()

<matplotlib.colorbar.Colorbar at 0x113a34760>
../../_images/7ec959834fc481038547a4169d0de177d9f01314e5dd33346edbc8bcef1ebfcb.png 
# Use meshgrid to create arrays (https://docs.scipy.org/doc/numpy/reference/generated/numpy.meshgrid.html)
mesh_b = np.ones((1,1000)) * np.linspace(-20, 20, 1000)
print(mesh_b.shape)
xx, yy = np.meshgrid(mesh, mesh_b)
print(xx.shape,yy.shape)
mesh_c = xx * yy
print(mesh_c.shape)

(1, 1000)
(1000, 1000) (1000, 1000)
(1000, 1000)

To better understand what meshgrid is doing, let’s plot xx, yy and mesh_c independently using subplots.

# plot xx, yy, mesh_c in 2-D
fig = plt.figure(figsize=(15,6))

# plot xx
plt.subplot(1,3,1)
plt.pcolormesh(xx)
plt.colorbar()

# plot yy
plt.subplot(1,3,2)
plt.pcolormesh(yy)
plt.colorbar()

# plot mesh_c
plt.subplot(1,3,3)
plt.pcolormesh(mesh_c)
plt.colorbar()

<matplotlib.colorbar.Colorbar at 0x113bfb820>
../../_images/b38ab76488aea9bc483437d3ea582452572f7d92e4a697acbb39c653cdc4c443.png 
# Plot the transpose
plt.pcolormesh(mesh.T)

<matplotlib.collections.QuadMesh at 0x113d1e200>
../../_images/b32f46ff0087c4006b7e96419900bc09c32152710fb151e96283cafbd6a0de34.png

There are many more operations you can do to arrays. Many of them are very intuitive, you don’t even need a reference :).

  • np.meshgrid(ndarray, ndarray)

  • np.reshape(ndarray, shape)

  • np.resize(ndarray, size)

  • np.tile(ndarray, reps)

  • ndarray.sum()

  • ndarray.mean()

  • ndarray.std()

  • plt.plot(x, y)

You can find the full NumPy reference here

We will take a look at some other plotting options in the next exercise.

Contents