Composing multi-panelled plots¶

One topic that people often find challenging when building plots with Python, is building plots with multiple panels. We'll have a look at how to do this using some of the automated options available in Seaborn, and then will look at how to apply more fine-grained control using Matplotlib and other libraries.

There are many different ways to approach building a multi-panel figure, and the best way to prepare yourself to be able to build any and all figures is to become used to reading the documentation. For every plot we make, find the relevant documentation page (in the API reference of the library) and compare the docstring to the way we use the functions.

Seaborn faceted plots¶

Seaborn is fantastic at quickly building multi-panelled plots. Lets loadd in our old friend, the penguin dataset, and start plotting.

In [1]:

# import libraries

import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
import numpy as np

# import libraries import matplotlib.pyplot as plt import seaborn as sns import pandas as pd import numpy as np

In [2]:

# load in our data
data = sns.load_dataset("penguins")

# load in our data data = sns.load_dataset("penguins")

Seaborn pairplot¶

The pairplot function is a figure level plot which will create a matrix of scatterplots of all the parameters, with a histogram down the central diagonal.

See documentation here

In [3]:

# using the defaults

sns.pairplot(data, hue="island")

# using the defaults sns.pairplot(data, hue="island")

Out[3]:

<seaborn.axisgrid.PairGrid at 0x7d5ca5a68a30>

In [4]:

#change the palette
sns.pairplot(data, hue="island", palette="husl")

#change the palette sns.pairplot(data, hue="island", palette="husl")

Out[4]:

<seaborn.axisgrid.PairGrid at 0x7d5ca0e03d00>

In [5]:

# select a subset of the data
sns.pairplot(
    data[["flipper_length_mm", "bill_length_mm", "body_mass_g", "island"]],
    hue="island", palette="Set2")

# select a subset of the data sns.pairplot( data[["flipper_length_mm", "bill_length_mm", "body_mass_g", "island"]], hue="island", palette="Set2")

Out[5]:

<seaborn.axisgrid.PairGrid at 0x7d5c9e671780>

Use a custom colour palette¶

In the introduction section, we discussed different places to find good colour palettes for scientific plotting, like ColorBrewer.

We can easily incorporate these palettes into our work by copying the hex code of the colours into a list.

In [6]:

custom_pal = ["#1b9e77", "#d95f02", "#7570b3"]

sns.pairplot(
    data,
    hue="island",
    palette=custom_pal)

custom_pal = ["#1b9e77", "#d95f02", "#7570b3"] sns.pairplot( data, hue="island", palette=custom_pal)

Out[6]:

<seaborn.axisgrid.PairGrid at 0x7d5c9db88a90>

In [8]:

# we can change the plot types being used with the keywords
# "kind" and "diag_kind"
# https://seaborn.pydata.org/generated/seaborn.pairplot.html

sns.pairplot(
    data[["flipper_length_mm", "bill_length_mm", "body_mass_g", "island"]],
    hue="island", palette="Set2",
    kind="reg", diag_kind="hist")

# we can change the plot types being used with the keywords # "kind" and "diag_kind" # https://seaborn.pydata.org/generated/seaborn.pairplot.html sns.pairplot( data[["flipper_length_mm", "bill_length_mm", "body_mass_g", "island"]], hue="island", palette="Set2", kind="reg", diag_kind="hist")

Out[8]:

<seaborn.axisgrid.PairGrid at 0x7d5c9c18cfa0>

In [10]:

# we can change the plot types being used with the keywords
# "kind" and "diag_kind"
# https://seaborn.pydata.org/generated/seaborn.pairplot.html

# we can pass keyword arguments to control the individual plot types


sns.pairplot(
    data[["flipper_length_mm", "bill_length_mm", "body_mass_g", "island"]],
    hue="island", palette="Set2",
    kind="reg", diag_kind="hist",
    diag_kws={"multiple":"stack"})

# we can change the plot types being used with the keywords # "kind" and "diag_kind" # https://seaborn.pydata.org/generated/seaborn.pairplot.html # we can pass keyword arguments to control the individual plot types sns.pairplot( data[["flipper_length_mm", "bill_length_mm", "body_mass_g", "island"]], hue="island", palette="Set2", kind="reg", diag_kind="hist", diag_kws={"multiple":"stack"})

Out[10]:

<seaborn.axisgrid.PairGrid at 0x7d5c9bb2fc10>

Seaborn jointplot¶

• Jointplot documentation

Again, this is a high-level figure function.

In [11]:

# we can call this plot in a single line
sns.jointplot(data, x="bill_length_mm", y="bill_depth_mm")

# we can call this plot in a single line sns.jointplot(data, x="bill_length_mm", y="bill_depth_mm")

Out[11]:

<seaborn.axisgrid.JointGrid at 0x7d5c9bda08e0>

In [12]:

# We can add further dimensions using "hue"
sns.jointplot(data, x="bill_length_mm", y="bill_depth_mm",
              hue="sex")

# note that seaborn automatically updates the type of marginal plot being shown

# We can add further dimensions using "hue" sns.jointplot(data, x="bill_length_mm", y="bill_depth_mm", hue="sex") # note that seaborn automatically updates the type of marginal plot being shown

Out[12]:

<seaborn.axisgrid.JointGrid at 0x7d5c9b417400>

In [15]:

# We can change the "kind" of plot
sns.jointplot(data, x="bill_length_mm", y="bill_depth_mm",
              hue="species", kind="kde")

# We can change the "kind" of plot sns.jointplot(data, x="bill_length_mm", y="bill_depth_mm", hue="species", kind="kde")

Out[15]:

<seaborn.axisgrid.JointGrid at 0x7d5c9b1380d0>

In [18]:

# we can add axis ticks to the marginal plot

sns.jointplot(data, x="bill_length_mm", y="bill_depth_mm", marginal_ticks=True)

# we can add axis ticks to the marginal plot sns.jointplot(data, x="bill_length_mm", y="bill_depth_mm", marginal_ticks=True)

Out[18]:

<seaborn.axisgrid.JointGrid at 0x7d5c9afdab90>

You can control the specific plots being shown in more detail using the "facet grid" function see documentation and examples here, but we are going to move on to look at more fine-grained control with basic matplotlib.

Matplotlib subplots¶

The subplot function is one of the most basic ways of creating a multi-panelled plot with fine-grained control.

Try out this gridded plot builder here

We'll start off using some very basic data, just to build a grid of plots.

In [19]:

# copied across from https://quick-subplots.streamlit.app/

# n.b. I often keep the ax titles set until I am finished making the plot,
# even if I don't want to use them in the fional plot, to help me locate
# which part of the grid I'm working on

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 2 rows and 2 columns
fig, axs = plt.subplots(2, 2, constrained_layout=True, sharex=False, sharey=False)

axs[0, 0].plot(x, y, 'tab:pink')
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

axs[1, 0].plot(x, y, 'tab:pink')
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

# copied across from https://quick-subplots.streamlit.app/ # n.b. I often keep the ax titles set until I am finished making the plot, # even if I don't want to use them in the fional plot, to help me locate # which part of the grid I'm working on # Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 2 rows and 2 columns fig, axs = plt.subplots(2, 2, constrained_layout=True, sharex=False, sharey=False) axs[0, 0].plot(x, y, 'tab:pink') axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') axs[1, 0].plot(x, y, 'tab:pink') axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

In [31]:

# what happens when we remove the following options:
# constrained_layout=True, sharex=False, sharey=False

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 2 rows and 2 columns
fig, axs = plt.subplots(2, 2,)

axs[0, 0].plot(x, y, 'tab:pink')
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

axs[1, 0].plot(x, y, 'tab:pink')
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

# what happens when we remove the following options: # constrained_layout=True, sharex=False, sharey=False # Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 2 rows and 2 columns fig, axs = plt.subplots(2, 2,) axs[0, 0].plot(x, y, 'tab:pink') axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') axs[1, 0].plot(x, y, 'tab:pink') axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

Out[31]:

[Text(0.5, 0, '1, 1 x-label'), Text(0, 0.5, '1, 1 y-label')]

In [33]:

# what happens when we remove the following options:
# constrained_layout=True, sharex=False, sharey=False

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 2 rows and 2 columns
fig, axs = plt.subplots(2, 2, constrained_layout=True, )

# we can create a scatter plot in one panel
axs[0, 0].scatter(data['bill_depth_mm'],
                  data['bill_length_mm'],
                  c='tab:pink', edgecolors="white", alpha=0.8)
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='Bill depth [mm]', ylabel='Bill length [mm]')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

axs[1, 0].plot(x, y, 'tab:pink')
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

# what happens when we remove the following options: # constrained_layout=True, sharex=False, sharey=False # Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 2 rows and 2 columns fig, axs = plt.subplots(2, 2, constrained_layout=True, ) # we can create a scatter plot in one panel axs[0, 0].scatter(data['bill_depth_mm'], data['bill_length_mm'], c='tab:pink', edgecolors="white", alpha=0.8) axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='Bill depth [mm]', ylabel='Bill length [mm]') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') axs[1, 0].plot(x, y, 'tab:pink') axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

Out[33]:

[Text(0.5, 0, '1, 1 x-label'), Text(0, 0.5, '1, 1 y-label')]

Let's add a histogram using the matplotlib function plt.hist - see the documentation here

In [35]:

# what happens when we remove the following options:
# constrained_layout=True, sharex=False, sharey=False

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 2 rows and 2 columns
fig, axs = plt.subplots(2, 2, constrained_layout=True, )

# we can create a scatter plot in one panel
axs[0, 0].scatter(data['bill_depth_mm'],
                  data['bill_length_mm'],
                  c='tab:pink', edgecolors="white", alpha=0.8)
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='Bill depth [mm]', ylabel='Bill length [mm]')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

# we can create a histogram in another panel
axs[1, 0].hist(data['bill_depth_mm'])
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='Bill depth [mm]', ylabel='Count')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

# what happens when we remove the following options: # constrained_layout=True, sharex=False, sharey=False # Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 2 rows and 2 columns fig, axs = plt.subplots(2, 2, constrained_layout=True, ) # we can create a scatter plot in one panel axs[0, 0].scatter(data['bill_depth_mm'], data['bill_length_mm'], c='tab:pink', edgecolors="white", alpha=0.8) axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='Bill depth [mm]', ylabel='Bill length [mm]') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') # we can create a histogram in another panel axs[1, 0].hist(data['bill_depth_mm']) axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='Bill depth [mm]', ylabel='Count') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

Out[35]:

[Text(0.5, 0, '1, 1 x-label'), Text(0, 0.5, '1, 1 y-label')]

In [38]:

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 2 rows and 2 columns
fig, axs = plt.subplots(2, 2, constrained_layout=True, )

# we can create a scatter plot in one panel
axs[0, 0].scatter(data['bill_depth_mm'],
                  data['bill_length_mm'],
                  c='tab:pink', edgecolors="white", alpha=0.8)
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='Bill depth [mm]', ylabel='Bill length [mm]')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

# we can create a histogram in another panel
axs[1, 0].hist(data['bill_depth_mm'],
               histtype="bar", color='tab:pink', edgecolor="white")
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='Bill depth [mm]', ylabel='Count')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

# Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 2 rows and 2 columns fig, axs = plt.subplots(2, 2, constrained_layout=True, ) # we can create a scatter plot in one panel axs[0, 0].scatter(data['bill_depth_mm'], data['bill_length_mm'], c='tab:pink', edgecolors="white", alpha=0.8) axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='Bill depth [mm]', ylabel='Bill length [mm]') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') # we can create a histogram in another panel axs[1, 0].hist(data['bill_depth_mm'], histtype="bar", color='tab:pink', edgecolor="white") axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='Bill depth [mm]', ylabel='Count') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

Out[38]:

[Text(0.5, 0, '1, 1 x-label'), Text(0, 0.5, '1, 1 y-label')]

In [40]:

# lets align the x-axis on the two left-hand plots

# can be done by setting limits, or by sharing the x axis

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 2 rows and 2 columns
fig, axs = plt.subplots(2, 2, constrained_layout=True, )

# we can create a scatter plot in one panel
axs[0, 0].scatter(data['bill_depth_mm'],
                  data['bill_length_mm'],
                  c='tab:pink', edgecolors="white", alpha=0.8)
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='Bill depth [mm]', ylabel='Bill length [mm]')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

# share axis with axs[0, 0]
axs[1, 0].sharex(axs[0,0])
# we can create a histogram in another panel
axs[1, 0].hist(data['bill_depth_mm'],
               histtype="bar", color='tab:pink', edgecolor="white",)
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='Bill depth [mm]', ylabel='Count')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

# lets align the x-axis on the two left-hand plots # can be done by setting limits, or by sharing the x axis # Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 2 rows and 2 columns fig, axs = plt.subplots(2, 2, constrained_layout=True, ) # we can create a scatter plot in one panel axs[0, 0].scatter(data['bill_depth_mm'], data['bill_length_mm'], c='tab:pink', edgecolors="white", alpha=0.8) axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='Bill depth [mm]', ylabel='Bill length [mm]') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') # share axis with axs[0, 0] axs[1, 0].sharex(axs[0,0]) # we can create a histogram in another panel axs[1, 0].hist(data['bill_depth_mm'], histtype="bar", color='tab:pink', edgecolor="white",) axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='Bill depth [mm]', ylabel='Count') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

Out[40]:

[Text(0.5, 0, '1, 1 x-label'), Text(0, 0.5, '1, 1 y-label')]

In [44]:

# lets align the x-axis on the two left-hand plots

# can be done by setting limits, or by sharing the x axis

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 2 rows and 2 columns
fig, axs = plt.subplots(2, 2, constrained_layout=True, )

# we can create a scatter plot in one panel
axs[0, 0].scatter(data['bill_depth_mm'],
                  data['bill_length_mm'],
                  c='tab:pink', edgecolors="white", alpha=0.8)
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='Bill depth [mm]', ylabel='Bill length [mm]')

# create another scatter plot sharing y axis
axs[0, 1].sharey(axs[0,0])
axs[0, 1].scatter(data['flipper_length_mm'],
               data['bill_length_mm'],
               c='tab:pink', alpha=0.5)
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='Flipper length [mm]', ylabel='Bill length [mm]')

# share axis with axs[0, 0]
axs[1, 0].sharex(axs[0,0])
# we can create a histogram in another panel
axs[1, 0].hist(data['bill_depth_mm'],
               histtype="bar", color='tab:pink', edgecolor="white",)
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='Bill depth [mm]', ylabel='Count')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

# lets align the x-axis on the two left-hand plots # can be done by setting limits, or by sharing the x axis # Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 2 rows and 2 columns fig, axs = plt.subplots(2, 2, constrained_layout=True, ) # we can create a scatter plot in one panel axs[0, 0].scatter(data['bill_depth_mm'], data['bill_length_mm'], c='tab:pink', edgecolors="white", alpha=0.8) axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='Bill depth [mm]', ylabel='Bill length [mm]') # create another scatter plot sharing y axis axs[0, 1].sharey(axs[0,0]) axs[0, 1].scatter(data['flipper_length_mm'], data['bill_length_mm'], c='tab:pink', alpha=0.5) axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='Flipper length [mm]', ylabel='Bill length [mm]') # share axis with axs[0, 0] axs[1, 0].sharex(axs[0,0]) # we can create a histogram in another panel axs[1, 0].hist(data['bill_depth_mm'], histtype="bar", color='tab:pink', edgecolor="white",) axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='Bill depth [mm]', ylabel='Count') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

Out[44]:

[Text(0.5, 0, '1, 1 x-label'), Text(0, 0.5, '1, 1 y-label')]

We can also use seaborn functions within the gridded layout.

We'll use the axes level functions available.

Let's start again with a fresh template.

In [45]:

# copied across from https://quick-subplots.streamlit.app/

# n.b. I often keep the ax titles set until I am finished making the plot,
# even if I don't want to use them in the fional plot, to help me locate
# which part of the grid I'm working on

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 2 rows and 2 columns
fig, axs = plt.subplots(2, 2, constrained_layout=True, sharex=False, sharey=False)

axs[0, 0].plot(x, y, 'tab:pink')
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

axs[1, 0].plot(x, y, 'tab:pink')
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

# copied across from https://quick-subplots.streamlit.app/ # n.b. I often keep the ax titles set until I am finished making the plot, # even if I don't want to use them in the fional plot, to help me locate # which part of the grid I'm working on # Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 2 rows and 2 columns fig, axs = plt.subplots(2, 2, constrained_layout=True, sharex=False, sharey=False) axs[0, 0].plot(x, y, 'tab:pink') axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') axs[1, 0].plot(x, y, 'tab:pink') axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

Out[45]:

[Text(0.5, 0, '1, 1 x-label'), Text(0, 0.5, '1, 1 y-label')]

In [47]:

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 2 rows and 2 columns
fig, axs = plt.subplots(2, 2, constrained_layout=True, sharex=False, sharey=False)

# replace this line:
# axs[0, 0].plot(x, y, 'tab:pink')
# with this:
sns.scatterplot(data, x='bill_depth_mm', y='bill_length_mm', ax=axs[0,0])
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

axs[1, 0].plot(x, y, 'tab:pink')
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

# Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 2 rows and 2 columns fig, axs = plt.subplots(2, 2, constrained_layout=True, sharex=False, sharey=False) # replace this line: # axs[0, 0].plot(x, y, 'tab:pink') # with this: sns.scatterplot(data, x='bill_depth_mm', y='bill_length_mm', ax=axs[0,0]) axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') axs[1, 0].plot(x, y, 'tab:pink') axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

Out[47]:

[Text(0.5, 0, '1, 1 x-label'), Text(0, 0.5, '1, 1 y-label')]

In [48]:

# What happens if you remove the x and y label settings?

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 2 rows and 2 columns
fig, axs = plt.subplots(2, 2, constrained_layout=True, sharex=False, sharey=False)

# replace this line:
# axs[0, 0].plot(x, y, 'tab:pink')
# with this:
sns.scatterplot(data, x='bill_depth_mm', y='bill_length_mm', ax=axs[0,0])
axs[0, 0].set_title('Axis [0, 0]')
# comment out this line
# axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

axs[1, 0].plot(x, y, 'tab:pink')
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

# What happens if you remove the x and y label settings? # Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 2 rows and 2 columns fig, axs = plt.subplots(2, 2, constrained_layout=True, sharex=False, sharey=False) # replace this line: # axs[0, 0].plot(x, y, 'tab:pink') # with this: sns.scatterplot(data, x='bill_depth_mm', y='bill_length_mm', ax=axs[0,0]) axs[0, 0].set_title('Axis [0, 0]') # comment out this line # axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') axs[1, 0].plot(x, y, 'tab:pink') axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

Out[48]:

[Text(0.5, 0, '1, 1 x-label'), Text(0, 0.5, '1, 1 y-label')]

What theme is being used by default when you plot with seaborn and matplotlib combined like this?

More advanced layout options¶

The matplotlib gridspec function provides very detailed control over the composition of our plots. Instead of using it directly, we are going to pass in some GridSpec keyword arguments to our subplots function (see here).

Once again, let's start with a fresh plot uysing this template, except let's give ourselves more panels to work with.

In [50]:

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 3 rows and 3 columns
fig, axs = plt.subplots(3, 3, constrained_layout=True, sharex=False, sharey=False)

axs[0, 0].plot(x, y, 'tab:pink')
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

axs[0, 2].plot(x, y, 'tab:pink')
axs[0, 2].set_title('Axis [0, 2]')
axs[0, 2].set(xlabel='0, 2 x-label', ylabel='0, 2 y-label')

axs[1, 0].plot(x, y, 'tab:pink')
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

axs[1, 2].plot(x, y, 'tab:pink')
axs[1, 2].set_title('Axis [1, 2]')
axs[1, 2].set(xlabel='1, 2 x-label', ylabel='1, 2 y-label')

axs[2, 0].plot(x, y, 'tab:pink')
axs[2, 0].set_title('Axis [2, 0]')
axs[2, 0].set(xlabel='2, 0 x-label', ylabel='2, 0 y-label')

axs[2, 1].plot(x, y, 'tab:pink')
axs[2, 1].set_title('Axis [2, 1]')
axs[2, 1].set(xlabel='2, 1 x-label', ylabel='2, 1 y-label')

axs[2, 2].plot(x, y, 'tab:pink')
axs[2, 2].set_title('Axis [2, 2]')
axs[2, 2].set(xlabel='2, 2 x-label', ylabel='2, 2 y-label')

# Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 3 rows and 3 columns fig, axs = plt.subplots(3, 3, constrained_layout=True, sharex=False, sharey=False) axs[0, 0].plot(x, y, 'tab:pink') axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') axs[0, 2].plot(x, y, 'tab:pink') axs[0, 2].set_title('Axis [0, 2]') axs[0, 2].set(xlabel='0, 2 x-label', ylabel='0, 2 y-label') axs[1, 0].plot(x, y, 'tab:pink') axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label') axs[1, 2].plot(x, y, 'tab:pink') axs[1, 2].set_title('Axis [1, 2]') axs[1, 2].set(xlabel='1, 2 x-label', ylabel='1, 2 y-label') axs[2, 0].plot(x, y, 'tab:pink') axs[2, 0].set_title('Axis [2, 0]') axs[2, 0].set(xlabel='2, 0 x-label', ylabel='2, 0 y-label') axs[2, 1].plot(x, y, 'tab:pink') axs[2, 1].set_title('Axis [2, 1]') axs[2, 1].set(xlabel='2, 1 x-label', ylabel='2, 1 y-label') axs[2, 2].plot(x, y, 'tab:pink') axs[2, 2].set_title('Axis [2, 2]') axs[2, 2].set(xlabel='2, 2 x-label', ylabel='2, 2 y-label')

Out[50]:

[Text(0.5, 0, '2, 2 x-label'), Text(0, 0.5, '2, 2 y-label')]

In [51]:

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 3 rows and 3 columns
# let's add gridspec arguments
fig, axs = plt.subplots(3, 3, constrained_layout=True, sharex=False, sharey=False,
                        gridspec_kw={
                            "width_ratios": [2, 4, 2]
                        })

axs[0, 0].plot(x, y, 'tab:pink')
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

axs[0, 2].plot(x, y, 'tab:pink')
axs[0, 2].set_title('Axis [0, 2]')
axs[0, 2].set(xlabel='0, 2 x-label', ylabel='0, 2 y-label')

axs[1, 0].plot(x, y, 'tab:pink')
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

axs[1, 2].plot(x, y, 'tab:pink')
axs[1, 2].set_title('Axis [1, 2]')
axs[1, 2].set(xlabel='1, 2 x-label', ylabel='1, 2 y-label')

axs[2, 0].plot(x, y, 'tab:pink')
axs[2, 0].set_title('Axis [2, 0]')
axs[2, 0].set(xlabel='2, 0 x-label', ylabel='2, 0 y-label')

axs[2, 1].plot(x, y, 'tab:pink')
axs[2, 1].set_title('Axis [2, 1]')
axs[2, 1].set(xlabel='2, 1 x-label', ylabel='2, 1 y-label')

axs[2, 2].plot(x, y, 'tab:pink')
axs[2, 2].set_title('Axis [2, 2]')
axs[2, 2].set(xlabel='2, 2 x-label', ylabel='2, 2 y-label')

# Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 3 rows and 3 columns # let's add gridspec arguments fig, axs = plt.subplots(3, 3, constrained_layout=True, sharex=False, sharey=False, gridspec_kw={ "width_ratios": [2, 4, 2] }) axs[0, 0].plot(x, y, 'tab:pink') axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') axs[0, 2].plot(x, y, 'tab:pink') axs[0, 2].set_title('Axis [0, 2]') axs[0, 2].set(xlabel='0, 2 x-label', ylabel='0, 2 y-label') axs[1, 0].plot(x, y, 'tab:pink') axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label') axs[1, 2].plot(x, y, 'tab:pink') axs[1, 2].set_title('Axis [1, 2]') axs[1, 2].set(xlabel='1, 2 x-label', ylabel='1, 2 y-label') axs[2, 0].plot(x, y, 'tab:pink') axs[2, 0].set_title('Axis [2, 0]') axs[2, 0].set(xlabel='2, 0 x-label', ylabel='2, 0 y-label') axs[2, 1].plot(x, y, 'tab:pink') axs[2, 1].set_title('Axis [2, 1]') axs[2, 1].set(xlabel='2, 1 x-label', ylabel='2, 1 y-label') axs[2, 2].plot(x, y, 'tab:pink') axs[2, 2].set_title('Axis [2, 2]') axs[2, 2].set(xlabel='2, 2 x-label', ylabel='2, 2 y-label')

Out[51]:

[Text(0.5, 0, '2, 2 x-label'), Text(0, 0.5, '2, 2 y-label')]

In [52]:

# Some example data to display
x = np.linspace(0, 2 * np.pi, 400)
y = np.sin(x ** 2)

# Set up figure with 3 rows and 3 columns
# let's add gridspec arguments
fig, axs = plt.subplots(3, 3, constrained_layout=True, sharex=False, sharey=False,
                        gridspec_kw={
                            "width_ratios": [2, 4, 2],
                            "height_ratios":[1, 3, 1]
                        })

axs[0, 0].plot(x, y, 'tab:pink')
axs[0, 0].set_title('Axis [0, 0]')
axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label')

axs[0, 1].plot(x, y, 'tab:pink')
axs[0, 1].set_title('Axis [0, 1]')
axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label')

axs[0, 2].plot(x, y, 'tab:pink')
axs[0, 2].set_title('Axis [0, 2]')
axs[0, 2].set(xlabel='0, 2 x-label', ylabel='0, 2 y-label')

axs[1, 0].plot(x, y, 'tab:pink')
axs[1, 0].set_title('Axis [1, 0]')
axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label')

axs[1, 1].plot(x, y, 'tab:pink')
axs[1, 1].set_title('Axis [1, 1]')
axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label')

axs[1, 2].plot(x, y, 'tab:pink')
axs[1, 2].set_title('Axis [1, 2]')
axs[1, 2].set(xlabel='1, 2 x-label', ylabel='1, 2 y-label')

axs[2, 0].plot(x, y, 'tab:pink')
axs[2, 0].set_title('Axis [2, 0]')
axs[2, 0].set(xlabel='2, 0 x-label', ylabel='2, 0 y-label')

axs[2, 1].plot(x, y, 'tab:pink')
axs[2, 1].set_title('Axis [2, 1]')
axs[2, 1].set(xlabel='2, 1 x-label', ylabel='2, 1 y-label')

axs[2, 2].plot(x, y, 'tab:pink')
axs[2, 2].set_title('Axis [2, 2]')
axs[2, 2].set(xlabel='2, 2 x-label', ylabel='2, 2 y-label')

# Some example data to display x = np.linspace(0, 2 * np.pi, 400) y = np.sin(x ** 2) # Set up figure with 3 rows and 3 columns # let's add gridspec arguments fig, axs = plt.subplots(3, 3, constrained_layout=True, sharex=False, sharey=False, gridspec_kw={ "width_ratios": [2, 4, 2], "height_ratios":[1, 3, 1] }) axs[0, 0].plot(x, y, 'tab:pink') axs[0, 0].set_title('Axis [0, 0]') axs[0, 0].set(xlabel='0, 0 x-label', ylabel='0, 0 y-label') axs[0, 1].plot(x, y, 'tab:pink') axs[0, 1].set_title('Axis [0, 1]') axs[0, 1].set(xlabel='0, 1 x-label', ylabel='0, 1 y-label') axs[0, 2].plot(x, y, 'tab:pink') axs[0, 2].set_title('Axis [0, 2]') axs[0, 2].set(xlabel='0, 2 x-label', ylabel='0, 2 y-label') axs[1, 0].plot(x, y, 'tab:pink') axs[1, 0].set_title('Axis [1, 0]') axs[1, 0].set(xlabel='1, 0 x-label', ylabel='1, 0 y-label') axs[1, 1].plot(x, y, 'tab:pink') axs[1, 1].set_title('Axis [1, 1]') axs[1, 1].set(xlabel='1, 1 x-label', ylabel='1, 1 y-label') axs[1, 2].plot(x, y, 'tab:pink') axs[1, 2].set_title('Axis [1, 2]') axs[1, 2].set(xlabel='1, 2 x-label', ylabel='1, 2 y-label') axs[2, 0].plot(x, y, 'tab:pink') axs[2, 0].set_title('Axis [2, 0]') axs[2, 0].set(xlabel='2, 0 x-label', ylabel='2, 0 y-label') axs[2, 1].plot(x, y, 'tab:pink') axs[2, 1].set_title('Axis [2, 1]') axs[2, 1].set(xlabel='2, 1 x-label', ylabel='2, 1 y-label') axs[2, 2].plot(x, y, 'tab:pink') axs[2, 2].set_title('Axis [2, 2]') axs[2, 2].set(xlabel='2, 2 x-label', ylabel='2, 2 y-label')

Out[52]:

[Text(0.5, 0, '2, 2 x-label'), Text(0, 0.5, '2, 2 y-label')]

For more fine-grained control, please see the GridSpec examples.

The Proplot library also has some nice multi-panelled gridding functionality.

The R/ggplot-inspired libraries have great functionality when it comes to faceting your data - see Lets-Plot and Plotnine.

In [ ]: