Combine choropleth map and barplot in Python

Choropleth maps combined with a bar chart are an excellent way of representing both the legend and the distribution of the underlying data. In this article, we'll create a step-by-step choropleth map with a bar chart to highlight unemployment in Belgium, using matplotlib.

About

A choropleth map is a map combined where colors are proportional to values in each region.

This chart has been created by Koen Van den Eeckhout, and translated to Python by Joseph Barbier.

Libraries

First, we need to load the following libraries:

import pandas as pd
import cartopy.crs as ccrs
import geopandas as gpd
import matplotlib.pyplot as plt
from pypalettes import add_cmap
from pyfonts import load_google_font
import unicodedata

Dataset

Here we need to load 2 datasets:

one with map data (shape of the Belgium). You can find it here.
one with data for each county. You can find it here.

And then we merge them.

def remove_accents(text):
    return "".join(
        c if not unicodedata.combining(c) else ""
        for c in unicodedata.normalize("NFKD", text)
    )


belgium = gpd.read_file("https://raw.githubusercontent.com/holtzy/The-Python-Graph-Gallery/refs/heads/master/static/data/belgium.json", layer="municipalities").drop(
    columns=[
        "prov_nis",
        "prov_fr",
        "prov_nl",
        "arr_nis",
        "id",
        "reg_nis",
        "reg_nl",
        "reg_fr",
        "arr_fr",
        "arr_nl",
        "nis",
    ]
)
belgium["name_nl"] = belgium["name_nl"].apply(remove_accents).str.lower()

rates = pd.read_csv("https://raw.githubusercontent.com/holtzy/The-Python-Graph-Gallery/refs/heads/master/static/data/belgium-unemployment.csv")
rates["Gemeente"] = rates["Gemeente"].apply(remove_accents).str.lower()

df = belgium.merge(rates, left_on="name_nl", right_on="Gemeente", how="left")

projection = ccrs.Mercator()
df.crs = "EPSG:4326"
df = df.to_crs(projection.proj4_init)
df.head()

	name_fr	name_nl	population	geometry	Gemeente	Werkloosheidsgraad
0	Anderlecht	anderlecht	120887	POLYGON ((476375.417 6554451.799, 476075.149 6...	anderlecht	15.72
1	Bruxelles	brussel	185103	POLYGON ((482638.162 6561047.697, 482766.849 6...	brussel	15.83
2	Ixelles	elsene	87632	MULTIPOLYGON (((489158.281 6555474.343, 489029...	elsene	11.87
3	Etterbeek	etterbeek	48473	POLYGON ((490488.042 6559495.241, 490445.146 6...	etterbeek	9.66
4	Evere	evere	42656	POLYGON ((492461.236 6563624.014, 492332.549 6...	evere	11.90

Basic choropleth map

The key steps here are:

create a colormap (cmap) with the color range we want
create a matplotlib Figure with fig, ax = plt.subplots()
plot the choropleth map with df.plot()

cmap = add_cmap(
    colors=[
        "#FCFCBD",
        "#FED69A",
        "#FDC48C",
        "#FB9A70",
        "#C74370",
        "#9D2D7A",
        "#86277A",
        "#661d5c",
        "#5A1A74",
    ],
    cmap_type="continuous",
    name="Sunset3",
)

fig, ax = plt.subplots(subplot_kw={"projection": projection}, dpi=300)
ax.axis("off")

df.plot(ax=ax, column="Werkloosheidsgraad", cmap=cmap, edgecolor="#e6e6e6", lw=0.3)
plt.show()

Add barplot

In order to add the barplot, we use the ax.inset_axes() function to create a subplot (smaller) that will contain our barplot.

If you're not familiar with complex layouts in Matplotlib, please check this dedicated lesson where we explain the concept in depth!

Then we customize it a bit so that it uses the right color scale and looks nice.

cmap = add_cmap(
    colors=[
        "#5A1A74",
        "#661d5c",
        "#86277A",
        "#9D2D7A",
        "#C74370",
        "#FB9A70",
        "#FDC48C",
        "#FED69A",
        "#FCFCBD",
    ][::-1],
    cmap_type="continuous",
    name="Sunset3",
)

fig, ax = plt.subplots(subplot_kw={"projection": projection}, dpi=300)
ax.axis("off")

df.plot(ax=ax, column="Werkloosheidsgraad", cmap=cmap, edgecolor="#e6e6e6", lw=0.3)

# Add barplot
bar_ax = ax.inset_axes(bounds=[0.05, 0.15, 0.4, 0.3], zorder=-1)
n, bins, _ = bar_ax.hist(df["Werkloosheidsgraad"], bins=18, alpha=0)
colors = [cmap((val - min(bins)) / (max(bins) - min(bins))) for val in bins]
bar_ax.bar(bins[:-1], n, color=colors)
bar_ax.spines[["top", "left", "right"]].set_visible(False)
bar_ax.set_yticks([])
x_ticks = list(range(0, 19, 3))
bar_ax.set_xticks(x_ticks, labels=["0", "3", "6", "9", "12", "15", "18%"], size=8)
bar_ax.tick_params(axis="x", length=2)

plt.show()

Add annotations

The final step is to add the annotations: title, subtitle and credit. The steps are:

load a regular and a bold font thanks to load_google_font() from pyfonts
use the fig.text() a few times to add the texts at the right places

regular = load_google_font("Roboto")
bold = load_google_font("Roboto", weight="bold")

cmap = add_cmap(
    colors=[
        "#5A1A74",
        "#661d5c",
        "#86277A",
        "#9D2D7A",
        "#C74370",
        "#FB9A70",
        "#FDC48C",
        "#FED69A",
        "#FCFCBD",
    ][::-1],
    cmap_type="continuous",
    name="Sunset3",
)

fig, ax = plt.subplots(subplot_kw={"projection": projection}, dpi=300)
ax.axis("off")

df.plot(
    ax=ax,
    column="Werkloosheidsgraad",
    cmap=cmap,
    edgecolor="#e6e6e6",
    lw=0.3,
)

bar_ax = ax.inset_axes(bounds=[0.05, 0.15, 0.4, 0.3], zorder=-1)
n, bins, _ = bar_ax.hist(df["Werkloosheidsgraad"], bins=18, alpha=0)
colors = [cmap((val - min(bins)) / (max(bins) - min(bins))) for val in bins]
bar_ax.bar(bins[:-1], n, color=colors)
bar_ax.spines[["top", "left", "right"]].set_visible(False)
bar_ax.set_yticks([])
x_ticks = list(range(0, 19, 3))
bar_ax.set_xticks(x_ticks, labels=["0", "3", "6", "9", "12", "15", "18%"], size=8)
bar_ax.tick_params(axis="x", length=2)

fig.text(x=0.2, y=0.89, s="Unemployment rate in Belgium", size=12, font=bold)
fig.text(x=0.2, y=0.86, s="By municipality, in December 2024", size=8, font=regular)
fig.text(
    x=0.2,
    y=0.13,
    s="Map: Koen Van den Eeckhout · Source: RVA (Interactive Statistics)",
    size=6,
    color="#909090",
    font=regular,
)

plt.savefig(
    "../../static/graph/web-choropleth-map-with-barplot.png",
    dpi=300,
    bbox_inches="tight",
)

Going further

🚨 Grab the Data To Viz poster!

Do you know all the chart types? Do you know which one you should pick? I made a decision tree that answers those questions. You can download it for free!