How can I make a blank subplot in matplotlib?

Question

I am making a group of subplot (say, 3 x 2) in matplotlib, but I have fewer than 6 datasets. How can I make the remaining subplot blank?

The arrangement looks like this:

+----+----+
| 0,0| 0,1|
+----+----+
| 1,0| 1,1|
+----+----+
| 2,0| 2,1|
+----+----+

This may go on for several pages, but on the final page, there are, for example, 5 datasets to the 2,1 box will be empty. However, I have declared the figure as:

cfig,ax = plt.subplots(3,2)

So in the space for subplot 2,1 there is a default set of axes with ticks and labels. How can I programatically render that space blank and devoid of axes?

Berkey

Answer 1

You could always hide the axes which you do not need. For example, the following code turns off the 6th axes completely:

import matplotlib.pyplot as plt

hf, ha = plt.subplots(3,2)
ha[-1, -1].axis('off')

plt.show()

and results in the following figure:

Alternatively, see the accepted answer to the question Hiding axis text in matplotlib plots for a way of keeping the axes but hiding all the axes decorations (e.g. the tick marks and labels).

Flyback

Thanks - that's actually even closer to my original question. I already accepted the other answer and adapted my code to use it, but both approaches are great. – Berkey

Cool, that's indeed nice as there's less add_subplot() clutter. – Philately

This is marvependous! – Pavlish

This does show no plot and one could say that this is a blank plot. I was looking for a plot which has no data, to explicitly show that there is no data. Any easy modification of this answer, which does that? – Curve

@Zelphir I'm not sure what you're after: the other five plots are empty graphs with no data, so I'm not sure what extra you're asking for. Also, this is a new question. In the future, please ask this as a new question rather than commenting on a old answer. – Flyback

@Chris, that last blank space is perfect for adding a legend. Do you know any way to set the legend there? – Coralloid

The answer is great but does not work for GeoAxes. Readers are also recommended using axes[-1,-1].set_visible(False) to meet the goal. – Millman

Answer 2

A much improved subplot interface has been added to matplotlib since this question was first asked. Here you can create exactly the subplots you need without hiding the extras. In addition, the subplots can span additional rows or columns.

import pylab as plt

ax1 = plt.subplot2grid((3,2),(0, 0))
ax2 = plt.subplot2grid((3,2),(0, 1))
ax3 = plt.subplot2grid((3,2),(1, 0))
ax4 = plt.subplot2grid((3,2),(1, 1))
ax5 = plt.subplot2grid((3,2),(2, 0))

plt.show()

Colloquium

Wow - that's a nice improvement. So much simpler! Thanks @Hooked! – Berkey

@Hooked, that last blank space is perfect for adding a legend. Do you know any way to set the legend there? – Coralloid

@Coralloid sure there are ways of doing this -- but this would be best posed as a new question, you can even link this answer in your question (welcome to StackOverflow btw!) – Colloquium

Answer 3

It's also possible to hide a subplot using the Axes.set_visible() method.

import matplotlib.pyplot as plt
import pandas as pd

fig = plt.figure()
data = pd.read_csv('sampledata.csv')

for i in range(0,6):
    ax = fig.add_subplot(3,2,i+1)
    ax.plot(range(1,6), data[i])
    if i == 5:
        ax.set_visible(False)

Rigamarole

Your answer is great for both original Axes and extended GeoAxes! – Millman

Answer 4

With newer versions of matplotlib (since 3.4.0), you can use an array of strings to design your subplots pattern (rows and columns) and use it to create the figure by calling matplotlib.pyplot.subplot_mosaic.

Instead of cfig, ax = plt.subplots(3,2), use cfig, axs = plt.subplot_mosaic(mosaic) and define mosaic this way:

mosaic = [['a', 'b'],
          ['c', 'd'],
          ['e', '.']]

In this pattern, blank subplots are denoted by '.' (by default, this can be parametrized in the call). You do not need to delete blank subplots, as they are not even created.

To select an axis for plotting, just use axs[id] where id is the string used to identify the subplot in the mosaic array.

Example:

mosaic = [['b', 'a'], ['.', 'au']]
kw = dict(layout='constrained')
fig, axs = plt.subplot_mosaic(mosaic, **kw)

ax = axs['b']
ax.grid(axis='y')
ax.bar(n, d)

ax = axs['a']
ax.grid(axis='y')
ax.bar(n, prior)

[...]

With subplot_mosaic, not only you can introduce blank subplots, but you can also merge 'cells' in order to create subplots on multiple rows and/or columns, just by crafting the required mosaic array, the rest of the code is unchanged. In addtion mosaic doesn't need to be an array, it can also be a multiline string. E.g. from Complex and semantic figure composition, using:

"""
A.C
BBB
.D.
"""

results in:

Stearic

So much more powerful! Thanks for the update @Stearic – Berkey

Answer 5

Would it be an option to create the subplots when you need them?

import matplotlib
matplotlib.use("pdf")
import matplotlib.pyplot as plt

plt.figure()
plt.gcf().add_subplot(421)
plt.fill([0,0,1,1],[0,1,1,0])
plt.gcf().add_subplot(422)
plt.fill([0,0,1,1],[0,1,1,0])
plt.gcf().add_subplot(423)
plt.fill([0,0,1,1],[0,1,1,0])
plt.suptitle("Figure Title")
plt.gcf().subplots_adjust(hspace=0.5,wspace=0.5)
plt.savefig("outfig")

Philately

I don't think so because there are other formatting things I need to do that I didn't include in the original question for brevity. One of these is plt.subplots_adjust(wspace=0,hspace=0). I'm not sure that would work after the fact. – Berkey

@Berkey : you can do subplots_adjust() using this approach. See my edit. – Philately

Answer 6

To delete the the plot positioned at (2,1) you may use

import matplotlib.pyplot as plt
cfig,ax = plt.subplots(3,2)
cfig.delaxes(ax.flatten()[5])

Blomquist