Improve your Matplotlib plots with this post

Rules for better figures

I found the defaults of Matplotlib fairly reasonable. Is fine for many purposes. But it kinda breaks many of the Ten simple rules for better figures¹. Just to mention one, the Rule 5: Do not trust the defaults. The main reason is to respect the other rules. In general, not following rule 5 works against Rule 1: Know Your Audience. In some other cases, it breaks Rule 8: Avoid “Chartjunk”. For example, if your data involves only two variables, you should have only two axes.

How I go about it

I have a custom style for plots that is more to my taste. The cover of this blog post has a comparison between two examples of the default Matplotlib style and my config.

Let’s take a closer look. The default Matplotlib style looks like this:

line_numbersUgly

If we apply my config, we can have it look like this:

line_numbers

Let’s learn how to set this style up.

Method A: The config file way

If you want more information about the Matplotlib config file, go here.

Set the style: This style is achieved with this config file

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21


font.family:       serif 
font.size:         12 

text.usetex:       True  # use LaTeX for all text handling.

axes.titlesize:    15    # fontsize of the axes title
axes.labelsize:    18    # fontsize of the x any y labels
axes.spines.top:   False # display axis spines
axes.spines.right: False

xtick.bottom:      True  # draw ticks on the bottom side
xtick.color:       black # color of the ticks
xtick.labelsize:   15    # fontsize of the tick labels

ytick.left:        True  # draw ticks on the left side
ytick.color:       black # color of the ticks
ytick.labelsize:   15    # fontsize of the tick labels

legend.fontsize:   12    # medium

pgf.rcfonts:       False

Save the config file on step 1 as myStyle.mplstyle. Once you load it all plots ran after it will have the custom style. You can load it with:

1
2


import matplotlib.pyplot as plt
plt.style.use('myStyle.mplstyle')

Alternative step 2

- For a local config: Save the file as matplotlibrc in the current directory.
- For a global config: Save the file as matplotlibrc to the directory shown by matplotlib.matplotlib_fname():

1
2
3


>>> import matplotlib
>>> matplotlib.matplotlib_fname() # show current matplotlibrc
'/home/foo/.config/matplotlib/matplotlibrc'

Method B: The dictionary way

There are other alternatives to achieve this style. The plt.style.use function accepts preferences as a dictionary:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19


import matplotlib.pyplot as plt
style_args = {
    "font.family":       "serif",
    "font.size":         12,
    "text.usetex":       True,
    "axes.titlesize":    15,
    "axes.labelsize":    18,
    "axes.spines.top":   False,
    "axes.spines.right": False,
    "xtick.bottom":      True,
    "xtick.color":       "black",
    "xtick.labelsize":   15,
    "ytick.left":        True,
    "ytick.color":       "black",
    "ytick.labelsize":   15,
    "legend.fontsize":   12,
    "pgf.rcfonts":       False,
}
plt.style.use(style_args)

Rougier NP, Droettboom M, Bourne PE (2014) Ten Simple Rules for Better Figures. PLoS Comput Biol 10(9): e1003833. https://doi.org/10.1371/journal.pcbi.1003833 ↩︎