Using external libraries and getting to grips with documentation¶

While matplotlib is often the first plotting library people encounter, it's far from the only one! Many libraries, such as seaborn, are build on top of matplotlib (meaning you can use matplotlib functions to change settings), while others are built from the ground up.

Some popular plotting libraries include:

• seaborn
• Plotly
• Vega Altair
• ProPlot
• Plotnine
• bokeh

Each of these uses slightly different function calls to create plots; and this can feel very overwhelming at first. However, it's also the dieal opportunity to get to grips with using Python documentaiton, because once you understand how to read the docs, implementing new cool libraries becomes an awful lot easier!

Building plots with seaborn¶

In the previous session, we used seaborn to set the style of our figure, but we were still using the matplotlb framework to actually create the plot. Lets have another look at what seaborn can do.

You can also look at our detailed course notes on building histograms here.

While we are working on these plots, keep the seaborn documentation open and keep checking back and forth how your work aligns with docstrings.

• seaborn histplots
• seaborn displots

We can create a bare-bones plot by setting up the fig and ax as we have done previously, then calling the seaborn histplot function. Unlike when we were using matplotlib, we don't prefix histplot with ax (like ax.plot); instead, we pass histplot the argument ax=. We've called our axes "ax", so we get the slightly silly looking argument ax=ax. We will see later in the course how this becomes more relevant when you have a multi-part figure!

In [1]:

# import necessary packages

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

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

In [2]:

# we're going to use the seaborn penguins dataset again

data = sns.load_dataset("penguins")

# we're going to use the seaborn penguins dataset again data = sns.load_dataset("penguins")

In [4]:

# we can use Colab's interactive data table tool to explore our data
data

# we can use Colab's interactive data table tool to explore our data data

Out[4]:

	species	island	bill_length_mm	bill_depth_mm	flipper_length_mm	body_mass_g	sex
0	Adelie	Torgersen	39.1	18.7	181.0	3750.0	Male
1	Adelie	Torgersen	39.5	17.4	186.0	3800.0	Female
2	Adelie	Torgersen	40.3	18.0	195.0	3250.0	Female
3	Adelie	Torgersen	NaN	NaN	NaN	NaN	NaN
4	Adelie	Torgersen	36.7	19.3	193.0	3450.0	Female
...	...	...	...	...	...	...	...
339	Gentoo	Biscoe	NaN	NaN	NaN	NaN	NaN
340	Gentoo	Biscoe	46.8	14.3	215.0	4850.0	Female
341	Gentoo	Biscoe	50.4	15.7	222.0	5750.0	Male
342	Gentoo	Biscoe	45.2	14.8	212.0	5200.0	Female
343	Gentoo	Biscoe	49.9	16.1	213.0	5400.0	Male

344 rows × 7 columns

In [7]:

# default seaborn theme
sns.set_theme()

# start the plot the same way as before: by defining the figure and axs objects
fig, ax = plt.subplots()

# now, we can use the seaborn histplot function
sns.histplot(data, x="bill_length_mm", ax=ax)

# default seaborn theme sns.set_theme() # start the plot the same way as before: by defining the figure and axs objects fig, ax = plt.subplots() # now, we can use the seaborn histplot function sns.histplot(data, x="bill_length_mm", ax=ax)

Out[7]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

In [10]:

# we can modify the theme
sns.set_theme(style="white", context="notebook",)

# start the plot the same way as before: by defining the figure and axs objects
fig, ax = plt.subplots()

# now, we can use the seaborn histplot function
sns.histplot(data, x="bill_length_mm", ax=ax)

# we can modify the theme sns.set_theme(style="white", context="notebook",) # start the plot the same way as before: by defining the figure and axs objects fig, ax = plt.subplots() # now, we can use the seaborn histplot function sns.histplot(data, x="bill_length_mm", ax=ax)

Out[10]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

In [12]:

# we can modify the theme
sns.set_theme(style="white", context="notebook",)

# start the plot the same way as before: by defining the figure and axs objects
fig, ax = plt.subplots()

# we can also define the data like this:
sns.histplot(data["bill_length_mm"], ax=ax)

# we can modify the theme sns.set_theme(style="white", context="notebook",) # start the plot the same way as before: by defining the figure and axs objects fig, ax = plt.subplots() # we can also define the data like this: sns.histplot(data["bill_length_mm"], ax=ax)

Out[12]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

In [14]:

# we can modify the theme
sns.set_theme(style="white", context="notebook",)

# start the plot the same way as before: by defining the figure and axs objects
fig, ax = plt.subplots()

# we can also define multiple distributions
sns.histplot(data[["bill_length_mm", "bill_depth_mm"]], ax=ax)

# we can modify the theme sns.set_theme(style="white", context="notebook",) # start the plot the same way as before: by defining the figure and axs objects fig, ax = plt.subplots() # we can also define multiple distributions sns.histplot(data[["bill_length_mm", "bill_depth_mm"]], ax=ax)

Out[14]:

<Axes: ylabel='Count'>

In [16]:

# we can modify the theme
sns.set_theme(style="white", context="notebook",)

# start the plot the same way as before: by defining the figure and axs objects
fig, ax = plt.subplots()

# we can add kde distributions
sns.histplot(data[["bill_length_mm", "bill_depth_mm"]], ax=ax, kde=True)

# we can modify the theme sns.set_theme(style="white", context="notebook",) # start the plot the same way as before: by defining the figure and axs objects fig, ax = plt.subplots() # we can add kde distributions sns.histplot(data[["bill_length_mm", "bill_depth_mm"]], ax=ax, kde=True)

Out[16]:

<Axes: ylabel='Count'>

In [15]:

# we can modify the theme
sns.set_theme(style="white", context="notebook",)

# start the plot the same way as before: by defining the figure and axs objects
fig, ax = plt.subplots()

# we can define "hue" to take one of the categorical parameters
sns.histplot(data, x="bill_length_mm", hue="species", ax=ax)

# we can modify the theme sns.set_theme(style="white", context="notebook",) # start the plot the same way as before: by defining the figure and axs objects fig, ax = plt.subplots() # we can define "hue" to take one of the categorical parameters sns.histplot(data, x="bill_length_mm", hue="species", ax=ax)

Out[15]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

In [17]:

# we can modify the theme
sns.set_theme(style="white", context="notebook",)

# start the plot the same way as before: by defining the figure and axs objects
fig, ax = plt.subplots()

# we can define "hue" to take one of the categorical parameters
# and can add the kde... until it's looking very busy!
sns.histplot(data, x="bill_length_mm", hue="species", ax=ax, kde=True)

# we can modify the theme sns.set_theme(style="white", context="notebook",) # start the plot the same way as before: by defining the figure and axs objects fig, ax = plt.subplots() # we can define "hue" to take one of the categorical parameters # and can add the kde... until it's looking very busy! sns.histplot(data, x="bill_length_mm", hue="species", ax=ax, kde=True)

Out[17]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

We are going to have a look at two different arguments that histplot can take that will help us control the appearance of our plot: multiple and element.

Argument	multiple	element
Default value	layer	bars
Options	“layer”, “dodge”, “stack”, “fill”	“bars”, “step”, “poly”

In [19]:

fig, ax = plt.subplots()
sns.histplot(data, x="bill_length_mm", hue="species", ax=ax, kde=True,
             multiple="stack", element="step")

fig, ax = plt.subplots() sns.histplot(data, x="bill_length_mm", hue="species", ax=ax, kde=True, multiple="stack", element="step")

Out[19]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

In [20]:

fig, ax = plt.subplots()
sns.histplot(data, x="bill_length_mm", hue="species", ax=ax, kde=True,
             multiple="dodge", element="bars")

fig, ax = plt.subplots() sns.histplot(data, x="bill_length_mm", hue="species", ax=ax, kde=True, multiple="dodge", element="bars")

Out[20]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

In [21]:

fig, ax = plt.subplots()
sns.histplot(data, x="bill_length_mm", hue="sex", ax=ax, kde=True,
             multiple="dodge", element="bars")

fig, ax = plt.subplots() sns.histplot(data, x="bill_length_mm", hue="sex", ax=ax, kde=True, multiple="dodge", element="bars")

Out[21]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

We can control the binwidth and use the shrink parameter to make plots like this more easy to read:

In [25]:

fig, ax = plt.subplots()
sns.histplot(data, x="bill_length_mm", hue="sex", ax=ax, kde=True,
             multiple="dodge", element="bars",
             binwidth=3, shrink=0.8)

fig, ax = plt.subplots() sns.histplot(data, x="bill_length_mm", hue="sex", ax=ax, kde=True, multiple="dodge", element="bars", binwidth=3, shrink=0.8)

Out[25]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

But sometimes, we need to split these apart and look at them separately.

For this, we can use the function displot(), which is a "figure-level wrapper" around histplot() that will do the work of organising different panels behind the scenes - much like the pyplot method of using matlpotlib.

In [38]:

sns.displot(data, x="bill_length_mm", row="sex", hue="species",
                  binwidth=3, multiple="stack", )

sns.displot(data, x="bill_length_mm", row="sex", hue="species", binwidth=3, multiple="stack", )

In [39]:

sns.displot(data, x="bill_length_mm", col="sex", hue="species",
                  binwidth=3, multiple="stack", )

sns.displot(data, x="bill_length_mm", col="sex", hue="species", binwidth=3, multiple="stack", )

Out[39]:

<seaborn.axisgrid.FacetGrid at 0x782edb9d7f40>

In [78]:

sns.set_theme(style="whitegrid", context="paper",)
sns.displot(data, x="bill_length_mm", col="sex", hue="species", row="island",
                  binwidth=3, multiple="stack", )

sns.set_theme(style="whitegrid", context="paper",) sns.displot(data, x="bill_length_mm", col="sex", hue="species", row="island", binwidth=3, multiple="stack", )

Out[78]:

<seaborn.axisgrid.FacetGrid at 0x782ed90b1de0>

However, much like with the pyplot ijnterface, there are certain disadvantages to working this way: you cannot control fine-level settings, such as figure size.

We'll look at how to manually control this in a later session.

We can also control the aethetics by changing the colour pallete:

In [45]:

sns.set_theme(palette="flare")

fig, ax = plt.subplots()
sns.histplot(data, x="bill_length_mm", hue="species", ax=ax, kde=True,
             multiple="stack", element="step",)

sns.set_theme(palette="flare") fig, ax = plt.subplots() sns.histplot(data, x="bill_length_mm", hue="species", ax=ax, kde=True, multiple="stack", element="step",)

Out[45]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

In [47]:

# we can improve the appearance by scaling the colour map better
pal = sns.color_palette("flare",3)

sns.set_theme(palette=pal, style="ticks")

fig, ax = plt.subplots()
sns.histplot(data, x="bill_length_mm", hue="species", ax=ax, kde=True,
             multiple="stack", element="step",)

# we can improve the appearance by scaling the colour map better pal = sns.color_palette("flare",3) sns.set_theme(palette=pal, style="ticks") fig, ax = plt.subplots() sns.histplot(data, x="bill_length_mm", hue="species", ax=ax, kde=True, multiple="stack", element="step",)

Out[47]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

In [53]:

# You can also set the palette on a plot-specific basis
# you can also edit bar edge colour and linewidth

sns.set_theme(style="white")

fig, ax = plt.subplots()
sns.histplot(data, x="bill_length_mm", hue="species", ax=ax,
             binwidth=1,
             multiple="stack",
             palette="light:m_r",
             edgecolor=".3",
             linewidth=.5,)

# You can also set the palette on a plot-specific basis # you can also edit bar edge colour and linewidth sns.set_theme(style="white") fig, ax = plt.subplots() sns.histplot(data, x="bill_length_mm", hue="species", ax=ax, binwidth=1, multiple="stack", palette="light:m_r", edgecolor=".3", linewidth=.5,)

Out[53]:

<Axes: xlabel='bill_length_mm', ylabel='Count'>

In [54]:

pal = sns.color_palette("flare",3)
sns.set_theme(palette=pal, style="ticks")

sns.displot(data, x="bill_length_mm", col="sex", hue="species",
                binwidth=1,
                multiple="stack",
                palette="light:m_r",
                edgecolor=".3",
                linewidth=.5,)

pal = sns.color_palette("flare",3) sns.set_theme(palette=pal, style="ticks") sns.displot(data, x="bill_length_mm", col="sex", hue="species", binwidth=1, multiple="stack", palette="light:m_r", edgecolor=".3", linewidth=.5,)

Out[54]:

<seaborn.axisgrid.FacetGrid at 0x782ee1b7a650>

What's an issue with the choice of colourmaps in these examples?

Recreating similar plots with different libraries¶

We discussed some other plotting libraries earlier; let's have a look at one of these now and see how it handles the task of building a histogram.

We are going to use the Plotly Express interface in the Plotly library - this is sort of similar to Matplotlib's pyplot interface, in that it is a simplified, high-level way of creating figures, but obscures more in-depth settings and functionality.

-Plotly express histogram docs

In [55]:

# import required libraries

import plotly.express as px

# import required libraries import plotly.express as px

In [61]:

fig = px.histogram(data, x="bill_length_mm")

# we need to include this for it to render:
fig.show()

fig = px.histogram(data, x="bill_length_mm") # we need to include this for it to render: fig.show()

In [63]:

# we can't specify bin width, but like seaborn, we can specify the number of bins
fig = px.histogram(data, x="bill_length_mm", nbins=100)
fig.show()

# we can't specify bin width, but like seaborn, we can specify the number of bins fig = px.histogram(data, x="bill_length_mm", nbins=100) fig.show()

In [65]:

# we can't specify bin width, but like seaborn, we can specify the number of bins
fig = px.histogram(data, x="bill_length_mm", nbins=100,
                   color="species")
fig.show()

# we can't specify bin width, but like seaborn, we can specify the number of bins fig = px.histogram(data, x="bill_length_mm", nbins=100, color="species") fig.show()

One of the key benefits of Plotly is it easily enables interactivity, without any effort or time spent designing a UI.

---

One final library we'll look at in this section is Plotnine. This library has a very different approach to the libraries we looked at previously: plotnine is inspired by the ggplot library for R and implements the same logic used in that library to develop it's plots.

Let's try to plot the same histogram using this library.

• Plotnine histogram

As in R, we need to map on aesthetic layers, and geometry layers (denotes aes and geom)

In [70]:

# import the library

import plotnine

# we're going to continue using our penguins dataset

# import the library import plotnine # we're going to continue using our penguins dataset

In [73]:

plotnine.ggplot(data,
                plotnine.aes(x="bill_length_mm")) + plotnine.geom_histogram()

plotnine.ggplot(data, plotnine.aes(x="bill_length_mm")) + plotnine.geom_histogram()

/usr/local/lib/python3.10/dist-packages/plotnine/stats/stat_bin.py:109: PlotnineWarning: 'stat_bin()' using 'bins = 11'. Pick better value with 'binwidth'.
/usr/local/lib/python3.10/dist-packages/plotnine/layer.py:284: PlotnineWarning: stat_bin : Removed 2 rows containing non-finite values.

This syntax might look somewhat familiar if you've previously used R, or it might be very unintuitive if you've only used Python.

We're not going to focus on this package today except to highlight its existence if you are more comfortable with R's "grammar of graphics".

In [75]:

plotnine.ggplot(data,
                plotnine.aes(
                    x="bill_length_mm")) + plotnine.geom_histogram(binwidth=1)

plotnine.ggplot(data, plotnine.aes( x="bill_length_mm")) + plotnine.geom_histogram(binwidth=1)

/usr/local/lib/python3.10/dist-packages/plotnine/layer.py:284: PlotnineWarning: stat_bin : Removed 2 rows containing non-finite values.

As with seaborn, Plotnine can be used to build a faceted grid. We'll look at more multi-panelled plots later in the course.

In [76]:

plotnine.ggplot(data, plotnine.aes(x="bill_length_mm", y=plotnine.after_stat("ncount"))) \
    + plotnine.geom_histogram(binwidth=0.5) \
    + plotnine.facet_grid("sex", "island")

plotnine.ggplot(data, plotnine.aes(x="bill_length_mm", y=plotnine.after_stat("ncount"))) \ + plotnine.geom_histogram(binwidth=0.5) \ + plotnine.facet_grid("sex", "island")

/usr/local/lib/python3.10/dist-packages/plotnine/layer.py:284: PlotnineWarning: stat_bin : Removed 2 rows containing non-finite values.

In [ ]: