Matplotlib tutorial (Advanced)

1.2. Matplotlib tutorial (Advanced)#

This tutorial continues on plotting with Python. Two packages will be introduced: Seaborn and Cartopy. Seaborn is a Python library for making statistical graphics based on Matplotlib and Cartopy is a Python package designed for geospatial data visualization. We will first illustrate alternative ways for plots in statistics with Seaborn, then show plotting of geospatial maps with Cartopy.

Plots in statistics with Seaborn: histograms, density plots, ecdf plots, bar charts, box plots, and scatter plots
Plotting maps with Cartopy: a simple example, adding features to the map, and plotting 2D data

1.2.1. Plots in statistics with Seaborn#

Seaborn helps you explore and understand your data. Its plotting functions operate on dataframes and arrays containing whole datasets and internally perform the necessary semantic mapping and statistical aggregation to produce informative plots. Its dataset-oriented, declarative API lets you focus on what the different elements of your plots mean, rather than on the details of how to draw them. Compared with Matplotlib, Seaborn helps you save your time and efforts on customizing the figures.

We first import the packages that might be used.

import seaborn as sns
import matplotlib as mpl
import matplotlib.pyplot as plt

Here, we utilize the “diamonds” dataset from Seaborn package for the demonstration of plots in statistics.

diamonds = sns.load_dataset("diamonds")
diamonds.head()

	carat	cut	color	clarity	depth	table	price	x	y	z
0	0.23	Ideal	E	SI2	61.5	55.0	326	3.95	3.98	2.43
1	0.21	Premium	E	SI1	59.8	61.0	326	3.89	3.84	2.31
2	0.23	Good	E	VS1	56.9	65.0	327	4.05	4.07	2.31
3	0.29	Premium	I	VS2	62.4	58.0	334	4.20	4.23	2.63
4	0.31	Good	J	SI2	63.3	58.0	335	4.34	4.35	2.75

1.2.1.1. Histograms#

There are two functions to plot histograms in Seaborn. displot provides figure-level interface for drawing distribution plots. By changing the argument kind, you can plot histograms, kernel density plots, and empirical cumulative distribution plots. The alternative functions are histplot(), kdeplot(), and ecdfplot().

# changes the global defaults
sns.set_theme(style="ticks")

f, ax = plt.subplots(figsize=(7, 5))

# removes the top and right axes spines
sns.despine(f)

# plot according to classification
sns.histplot(
    diamonds,
    x="price", hue="cut",
    multiple="stack",
    palette="light:m_r",
    edgecolor=".3",
    linewidth=.5,
    log_scale=True,  # log scale
)
ax.xaxis.set_major_formatter(mpl.ticker.ScalarFormatter())
ax.set_xticks([500, 1000, 2000, 5000, 10000])

[<matplotlib.axis.XTick at 0x2bdfb39c190>,
 <matplotlib.axis.XTick at 0x2bdfc8e73a0>,
 <matplotlib.axis.XTick at 0x2bdfc7b7e20>,
 <matplotlib.axis.XTick at 0x2bdfc7bcdf0>,
 <matplotlib.axis.XTick at 0x2bdfcab6b50>]

../../_images/7444ff52e97d0a45a8eefca00cf35498153e118c0895ea6c406f2ab19267a200.png

Alternatively, you can use displot() function. The other settings are more or less similar.

# changes the global defaults
sns.set_theme(style="ticks")

# removes the top and right axes spines
sns.despine(f)

# plot according to classification
sns.displot(
    diamonds,
    x="price", hue="cut",
    multiple="stack",
    palette="light:m_r",
    edgecolor=".3",
    linewidth=.5,
    log_scale=True,  # log scale
    kind='hist'
)

[<matplotlib.axis.XTick at 0x2bd828d8670>,
 <matplotlib.axis.XTick at 0x2bd828d8640>,
 <matplotlib.axis.XTick at 0x2bd828dec10>,
 <matplotlib.axis.XTick at 0x2bd80014f70>,
 <matplotlib.axis.XTick at 0x2bd80014130>]

../../_images/b8feadfa2f6810a9080d8769c1a75ad1a50753d539e6d94799fcff9c995ee79b.png

1.2.1.2. Density plots#

# changes the global defaults
sns.set_theme(style="ticks")

f, ax = plt.subplots(figsize=(7, 5))

# removes the top and right axes spines
sns.despine(f)

# plot according to classification
sns.kdeplot(
    data=diamonds,
    x="price", hue="cut",multiple="stack",
    fill=True, common_norm=True,  palette="crest",
    alpha=.5, 
    log_scale=True,  # log scale
)
ax.xaxis.set_major_formatter(mpl.ticker.ScalarFormatter())
ax.set_xticks([500, 1000, 2000, 5000, 10000])

[<matplotlib.axis.XTick at 0x2bdfd0cd2b0>,
 <matplotlib.axis.XTick at 0x2bdfd0cd280>,
 <matplotlib.axis.XTick at 0x2bdfd0ca430>,
 <matplotlib.axis.XTick at 0x2bdfd0f86a0>,
 <matplotlib.axis.XTick at 0x2bdfd0f8bb0>]

../../_images/d90aadc4237629fc43d9e5f9d2288465905981674902f02618f2100f69cbde6b.png

Alternatively,

# changes the global defaults
sns.set_theme(style="ticks")

# removes the top and right axes spines
sns.despine(f)

# plot according to classification
sns.displot(
    data=diamonds,
    x="price", hue="cut",multiple="stack",
    fill=True, common_norm=True,  palette="crest",
    alpha=.5, 
    log_scale=True,  # log scale
    kind='kde'
)

<seaborn.axisgrid.FacetGrid at 0x2bd80ff94f0>

../../_images/ca49860963426cac9293ebb412d03f4b46f7a3dae0c061cd842c3699b6153f1d.png

1.2.1.3. Empirical cumulative distribution functions (ecdf) plots#

An ECDF represents the proportion or count of observations falling below each unique value in a dataset. It also aids direct comparisons between multiple distributions.

f, ax = plt.subplots(figsize=(7, 5))

# plot according to classification
sns.ecdfplot(
    data=diamonds,
    x="price", hue="cut",
    palette="crest",
    alpha=.8, 
)

<AxesSubplot:xlabel='price', ylabel='Proportion'>

../../_images/d88b93ca69b934ef976bc8966ed834a1b782a926f9750e05cfe53e746246ecc8.png

# plot according to classification
sns.displot(
    data=diamonds,
    x="price", hue="cut",
    palette="crest",
    alpha=.8, 
    kind='ecdf'
)

<seaborn.axisgrid.FacetGrid at 0x2bd8791c2b0>

../../_images/f9a0a0f1c6e0a074e26be7429a8ff0a8e92e03204f709f8ef2a5eb1ac0a98e24.png

1.2.1.4. Bar charts#

Bar charts are one of the categorical plots, which could be handled by catplot() with argument kind="bar". Besides, catplot() can also generate box plots, violin plots, swarm plots, and so on.

g = sns.catplot(
    data=diamonds, kind="bar",
    x="color", y="price", 
    ci=99, palette="dark", alpha=.6, height=6
)
# ci: Size of confidence intervals to draw around estimated values. 

g.despine(left=True)
g.set_axis_labels("Color", "Price")

<seaborn.axisgrid.FacetGrid at 0x2bd81649730>

../../_images/53277afd836beb384b24f929bc8df269771247ca9fdeefb54b980726d477f025.png

The other way for bar charts is barplot() function.

f, ax = plt.subplots(figsize=(7, 5))

sns.barplot(
    data=diamonds, 
    x="color", y="price", 
    ci=99, palette="dark", alpha=.6
)
# ci: Size of confidence intervals to draw around estimated values. 

sns.despine(f)

../../_images/7aa289f0356a1982ec509647c1ef61b4baf81e8576f1bf7a98c6a8fa02882b29.png

1.2.1.5. Box plots#

The box plot is another kind of categorical plots to show the distribution of the data.

g = sns.catplot(
    data=diamonds, kind="box",
    x="cut", y="price", 
    palette="dark", 
    height=6,
    orient="v",
    whis=2,
    fliersize=0.5
)
# fliersize: Size of the markers used to indicate outlier observations.
# whis: Proportion of the IQR past the low and high quartiles to extend the plot whiskers.

g.despine(left=True)
g.set_axis_labels("Cut", "Price")

<seaborn.axisgrid.FacetGrid at 0x2bd80377ac0>

../../_images/b0ab7bb865e2844c9f882d912e48e94dfa42021b5837a5e54843c15f0d0490d2.png

Alternatively,

f, ax = plt.subplots(figsize=(7, 5))

sns.boxplot(
    data=diamonds, 
    x="cut", y="price", 
    palette="dark", 
    orient="v",
    whis=2,
    fliersize=0.5
)
# fliersize: Size of the markers used to indicate outlier observations.
# whis: Proportion of the IQR past the low and high quartiles to extend the plot whiskers.

sns.despine(f)

../../_images/3f87688a3ec179bed5c80c2939a7b11849d5979391d9fbe251806443875f01a7.png

1.2.1.6. Scatter plots#

Scatter plots could be generated by scatterplot(). For example,

sns.scatterplot(data=diamonds, x="x", y="price", alpha=0.1, edgecolor=None)

<AxesSubplot:xlabel='x', ylabel='price'>

../../_images/97ebed5f1bd61f02ce65bfc683f6ab44f026f022254fedc93de4a11469906bc8.png

In addition, Seaborn has a function called pairplot(), which shows the scatter matrix and is extremely useful for multivariate analysis. We use “penguins” dataset for this example.

df = sns.load_dataset("penguins")
sns.pairplot(df, hue="species", markers=["o", "s", "D"])

<seaborn.axisgrid.PairGrid at 0x2bd8d717040>

../../_images/cd1dd917f7b0d1f4d1fdc7b5db03f630de56eba4bb3b7982c2dff3af9abb7786.png

1.2.3. References#

Seaborn documentation
Cartopy documentation
Tutorial on plotting maps with python by Julius Busecke
Research computing in Earth Sciences - Maps with Cartopy
ERA5-Land monthly averaged data from 1981 to present

Matplotlib tutorial (Advanced)

Contents

1.2. Matplotlib tutorial (Advanced)#

1.2.1. Plots in statistics with Seaborn#

1.2.1.1. Histograms#

1.2.1.2. Density plots#

1.2.1.3. Empirical cumulative distribution functions (ecdf) plots#

1.2.1.4. Bar charts#

1.2.1.5. Box plots#

1.2.1.6. Scatter plots#

1.2.2. Plotting maps with Cartopy#

1.2.2.1. A simple example#

1.2.2.2. Adding features to the map#

1.2.2.3. Plotting 2D data#

1.2.3. References#