Matplotlib - 将线添加到链接堆叠条 plot 类别

Question

I'm able to generate stacked bar plot.

But how to generate something like this one?

I have two sets of data per country. This two sets (power and CO2) have not the same unit and are represented today on two different graphs. It would be more readable to present them on the same one.

I have no clue which properties of the bar chart could help me to do this.

Answer 1

A bit of a late response, but hopefully this reply can help someone out who might be stumbling onto the post with the same or similar question.

Replicating the example graph provided by OP with paired data for each order along the x axis, I came up with the following solution:

import matplotlib.pyplot as plt
import numpy as np

order = ["First", "Second", "Third", "Fourth"]
order_axis = np.arange(len(order))

a_main = np.array([15, 20, 16, 18])
b_main = np.array([10, 12, 5, 9])

a_other = np.array([14, 10, 8, 16])
b_other = np.array([17, 19, 4, 11])

w = 0.2
sep = 0.1

for axis in order_axis:
    plt.plot([order_axis[axis] - sep, order_axis[axis] + sep], [a_main[axis], a_other[axis]], color='C7', ls='--', zorder=1)
    plt.plot([order_axis[axis] - sep, order_axis[axis] + sep], [a_main[axis] + b_main[axis], a_other[axis] + b_other[axis]], color='C7', ls='--', zorder=1)

plt.bar(order_axis - sep, a_main, -w, color='C0', zorder=2, align="edge")
plt.bar(order_axis - sep, b_main, -w, color='C1', bottom=a_main, zorder=2, align="edge")

plt.bar(order_axis + sep, a_other, w, color='C0', zorder=2, align="edge")
plt.bar(order_axis + sep, b_other, w, color='C1', bottom=a_other, zorder=2, align="edge")

plt.xticks(order_axis, order)
plt.show()

This is the graph produced by the code above

For good measure, I also have a solution that works generally for any number of stacked bars with any number of orders along the x axis.

order = ["First", "Second", "Third", "Fourth", "Fifth"]

a = np.array([17, 15, 16, 18, 10])
b = np.array([8, 12, 5, 9, 15])
c = np.array([6, 3, 6, 4, 3])

fig = plt.figure()
ax = plt.subplot(111)

ax.bar(order, a, color='C0', zorder=2)
ax.bar(order, b, color='C1', bottom=a, zorder=2)
ax.bar(order, c, color='C2', bottom=a+b, zorder=2)

width = ax.patches[0].get_width()
stacks = len(ax.patches) // len(order)

for i in range(stacks):
    for j in range(0, len(order) - 1):
        h0 = np.sum([ax.patches[j + len(order) * k].get_height() for k in range(0, i + 1)])
        h1 = np.sum([ax.patches[j + 1 + len(order) * k].get_height() for k in range(0, i + 1)])

        ax.plot([j + width / 2, j + 1 - width / 2], [h0, h1], color='C7', ls='--', zorder=1)

plt.show()

Graph produced by the second code snippet