Scatter plots using matplotlib.pyplot.scatter()

Scatter plots using matplotlib.pyplot.scatter()

First, let’s install pyplot from matplotlib and call it plt:

import matplotlib.pyplot as plt

We are also going to need some data which we’ll create using numpy - type the following:

import numpy as np

Now lets create some random point data to mimic some xy coordinates and some associated attribute:


The basic scatter

To create our plot, we are going to use the plt.scatter() function (remember to check out the function help by using plt.scatter?) - an alternative to plt.plot() which gives you more control on setting colours based on another variable. This function takes in 2 variables to plot - we’ll use the first 2 columns of our xyz array:

plt.scatter(xyz[:,0], xyz[:,1])

You should see something like the following being printed out:

>>> <matplotlib.collections.PathCollection at 0x1afd1d30>

To view the image, we can use the function - type this in:

and you should see something like this:

"Your first matplotlib scatter plot"

Improving appearance

This is all well and good but we are missing some important components - for example axis labels and a title. These are easily added - first you must re-create the scatter plot:

plt.scatter(xyz[:,0], xyz[:,1])

Using the created plt instance, you can add labels like this:

plt.title("Point observations")

If you’ve had a look at the documentation for plt.scatter() you will also see that the function can take in a scalar to adjust the marker size (starts at a default value of 20). To make these smaller, you must pass in a value to the plt.scatter() function:

plt.scatter(xyz[:,0], xyz[:,1], marker_size)

What we can also do is change the colour of the points - want them in red? Then when invoking plt.scatter(), you’ll need to set the c flag (more info on colours can be found here):

plt.scatter(xyz[:,0], xyz[:,1], c='r')

Something else that can be handy is to colour the points by another variable - in the case of our xyz test data, currently we are only using the first 2 columns - lets use the third colmn to colour the plot:

plt.scatter(xyz[:,0], xyz[:,1], c=xyz[:,2])

By adding these new colours, we now have information on the plot that alone is not particularly informative - we need a colorbar and fortunately, there is a method for creating this - plt.colorbar():

plt.scatter(xyz[:,0], xyz[:,1], c=xyz[:,2])

Last but not least, lets add a title to the colorbar to indicate what it represents - to do this, after creating your initial plot, assign the creation of your colorbar to a variable like this:

cbar = plt.colorbar()

You can now access methods of colorbar - have a look at what’s available by typing cbar. followed by the tab key.

To set the title of the colorbar, we need to type:

cbar.set_label("elevation (m)")

You can change the distance the label is from the colorbar by using the labelpad option (positive moves away, negative moves it closer):

cbar.set_label("elevation (m)", labelpad=-1)

Now, to create your final plot just type:

plt.scatter(xyz[:,0], xyz[:,1], marker_size, c=xyz[:,2])
plt.title("Point observations")
cbar= plt.colorbar()
cbar.set_label("elevation (m)", labelpad=+1)

You should end up with something like this:

"Your finalised scatter plot"

Note - if you have created multiple scatter plots (i.e. have enterered plt.scatter() a few times with no call to, then these will all be plotted visually on your call to If you are concerned that this is going to happen (and you only want to display your most recent plt.scatter() call), then type plt.clf() which clears everything, and then retype the code to create your figure.

Again, to be safe and ensure everything is clean, type plt.clf() again.

Previous Home Next