參考資料：
https://github.com/mikhailsirenko/bivariate-choropleth/blob/main/bivariate-choropleth.ipynb
Bivariate choropleth map using Plotly

Matplotlib雙變量熱力等級圖

代碼：

import pandas as pd
import geopandas as gpd
import numpy as np
import matplotlib.pyplot as plt
import matplotlib
from adjustText import adjust_text
import contextily as ctx
from sklearn.preprocessing import MinMaxScaler
from scipy.stats.mstats import winsorizedata = gpd.read_file('data.json')
data.head()
border = gpd.read_file('border.json')# Define the bins
bins = [0, 0.33, 0.66, 1]# Bin the first variable - x
data['Var1_Class'] = pd.cut(data['x'], bins=bins, include_lowest=True)
data['Var1_Class'] = data['Var1_Class'].astype('str')# Bin the second variable - y
data['Var2_Class'] = pd.cut(data['y'], bins=bins, include_lowest=True)
data['Var2_Class'] = data['Var2_Class'].astype('str')# Code created x bins to 1, 2, 3
x_class_codes = np.arange(1, len(bins))
d = dict(zip(data['Var1_Class'].value_counts().sort_index().index, x_class_codes))
data['Var1_Class'] = data['Var1_Class'].replace(d)# Code created y bins to A, B, C
y_class_codes = ['A', 'B', 'C']
d = dict(zip(data['Var2_Class'].value_counts().sort_index().index, y_class_codes))
data['Var2_Class'] = data['Var2_Class'].replace(d)# Combine x and y codes to create Bi_Class
data['Bi_Class'] = data['Var1_Class'].astype('str') + data['Var2_Class']all_colors = ['#e8e8e8', '#b0d5df', '#64acbe', '#e4acac', '#ad9ea5', '#627f8c', '#c85a5a', '#985356', '#574249']
colors = ['#e8e8e8', # 1A'#b0d5df', # 1B# '#64acbe', # 1C'#e4acac', # 2A'#ad9ea5', # 2B# '#627f8c', # 2C'#c85a5a', # 3A'#985356'] # 3B# '#574249'] # 3Ccmap = matplotlib.colors.ListedColormap(colors)
fig, ax = plt.subplots(figsize=(8,8))# Step 1: Draw the map
border.to_crs('EPSG:3857').plot(ax=ax, facecolor='none', edgecolor='black', alpha=.5) # city border
data.to_crs('EPSG:3857').plot(ax=ax, column='Bi_Class', # variable that is going to be used to color the map cmap=cmap, # newly defined bivariate cmap categorical=True, # bivariate choropleth has to be colored as categorical map legend=False) # we're going to draw the legend ourselves
ctx.add_basemap(ax=ax, source=ctx.providers.CartoDB.Positron) # add the basemapplt.tight_layout() # "tighten" two figures map and basemap
plt.axis('off') # we don't need axis with coordinates
ax.set_title('Bivariate Choropleth Amsterdam')# Step 2: draw the legend# We're drawing a 3x3 "box" as 3 columns
# The xmin and xmax arguments axvspan are defined to create equally sized small boxesimg2 = fig # refer to the main figure
ax2 = fig.add_axes([0.15, 0.25, 0.1, 0.1]) # add new axes to place the legend there# and specify its location 
alpha = 1 # alpha argument to make it more/less transperent# Column 1
ax2.axvspan(xmin=0, xmax=0.33, ymin=0, ymax=0.33, alpha=alpha, color=all_colors[0])
ax2.axvspan(xmin=0, xmax=0.33, ymin=0.33, ymax=0.66, alpha=alpha, color=all_colors[1])
ax2.axvspan(xmin=0, xmax=0.33, ymin=0.66, ymax=1, alpha=alpha, color=all_colors[2])# Column 2
ax2.axvspan(xmin=0.33, xmax=0.66, ymin=0, ymax=0.33, alpha=alpha, color=all_colors[3])
ax2.axvspan(xmin=0.33, xmax=0.66, ymin=0.33, ymax=0.66, alpha=alpha, color=all_colors[4])
ax2.axvspan(xmin=0.33, xmax=0.66, ymin=0.66, ymax=1, alpha=alpha, color=all_colors[5])# Column 3
ax2.axvspan(xmin=0.66, xmax=1, ymin=0, ymax=0.33, alpha=alpha, color=all_colors[6])
ax2.axvspan(xmin=0.66, xmax=1, ymin=0.33, ymax=0.66, alpha=alpha, color=all_colors[7])
ax2.axvspan(xmin=0.66, xmax=1, ymin=0.66, ymax=1, alpha=alpha, color=all_colors[8])# Step 3: annoate the legend
ax2.tick_params(axis='both', which='both', length=0) # remove ticks from the big box
ax2.axis('off'); # turn off its axis
ax2.annotate("", xy=(0, 1), xytext=(0, 0), arrowprops=dict(arrowstyle="->", lw=1)) # draw arrow for x 
ax2.annotate("", xy=(1, 0), xytext=(0, 0), arrowprops=dict(arrowstyle="->", lw=1)) # draw arrow for y 
ax2.text(s='Var1\_Class', x=0.1, y=-0.25) # annotate x axis
ax2.text(s='Var2\_Class', x=-0.25, y=0.1, rotation=90); # annotate y axis
plt.savefig('bivariate_choropleth.png', dpi=300)

代碼來源：https://github.com/mikhailsirenko/bivariate-choropleth/blob/main/bivariate-choropleth.ipynb