Matplotlib | Donut graph! Percentage % in the center (pie, text)

A pie chart is often used to display data proportions, but for a more modern design, a doughnut graph is sometimes used by hollowing out the center of the pie chart.

This article explains how to plot donut and double donut graphs in Matplotlib

Detailed instructions are also provided on how to customize the donut graph legend, labels, percentages, changing element coordinates, colors, colormaps, thickness, text, and more.

Please refer to the following article regarding pie charts

あわせて読みたい

Matplotlib | Pie chart! Legend, Percentage, Labels (pie) Pie charts are often used to show proportions of data. This article explains how to plot a pie chart in Matplotlib. In addition, detailed instructions are pr…

Options for doughnut graphs

In Matplotlib, donut graphs are drawn with the Axes.pie function.

Basic donut graph (wedgeprops, color)

Draws a donut graph with the center portion hollowed out from a pie chart.

The donut graph is drawn by changing the width of the pie chart using the wedgeprops argument of the Axes.pie function.

Width of the elements (wedgeprops)

The argument wedgeprops is specified as a dictionary type, {'width': 0.3}.

The following tabs explain the code and flowchart

import matplotlib.pyplot as plt
import numpy as np

# step1 Create data
width = 0.3
vals = [40, 60]

# step2 Create graph frames
fig, ax = plt.subplots()

# step3 Plot a donut chart
ax.pie(vals, startangle=90,
        # Pie Chart Width
        wedgeprops={'width':width}
)
ax.set_title('Donut chart')

plt.show()

Colors of the elements (colors)

Colors of the elements are specified by colors=array.

We used a color gradient with a colormap called Blues.

import matplotlib.pyplot as plt
import numpy as np

# step1 Create data
width = 0.3
vals = [40, 60]
# step1.1 Create a colormap
cmap = plt.get_cmap('Blues')
colors = cmap(np.linspace(0.2, 0.7, len(vals)))

# step2 Create graph frames
fig, ax = plt.subplots()

# step3 Plot a donut chart
ax.pie(vals, startangle=90,
        # Donut Chart Width
        wedgeprops={'width':width},
        # Add colors
        colors=colors
)
ax.set_title('Donut chart')
plt.show()

Percentage of circle center (Axes.text)

The Axes.text function displays a percentage % or strings in the center of the doughnut graph.

The following is a list of options for the Axes.text function

Show Percentage % (fontsize)

Specify coordinates (x, y) = (0, 0) and strings (s) = 60% + Donut Chart as arguments to Axes.text.

In addition, we set font size = 20

# step4 Add text
ax.text(0, 0, r'60%'+'\nDonut Chart', fontsize=20)

Place a percentage % (horizontalalignment, verticalalignment)

In the Show Percentage %, the position of the displayed text was not centered.

horizontalalignment='center' for horizontal, verticalalignment='center' for vertical

It can also be abbreviated as horizontalalignment→ha, verticalalignment→va.

# step4 Add text
ax.text(0, 0, r'60%'+'\nDonut Chart', fontsize=20, 
        horizontalalignment='center', 
        verticalalignment='center'
)

Modern design donut graph (color)

Separate the percentage from the rest of the text and highlight only the percentage.

First, use two Axes.text functions, one for the percentage text you want to emphasize and one for the rest of the text.

Utilizing the verticalalignment of Place a Percentage, the top text is set to bottom and the bottom text to top, there is no interference at any font size.

# step1.1 Create a colormap
color_text = cmap(0.7)

# step4 Add text
ax.text(0, 0, r'60%', color=color_text, fontsize=32, fontweight='medium',
        horizontalalignment='center', verticalalignment='bottom')
ax.text(0, 0, r'Donut Chart', color=color_text, fontsize=20, 
        horizontalalignment='center', verticalalignment='top')

Legend and numpy.array (Axes.legend)

This section explains how to display and arrange a legend on a donut graph.

In addition, the ndarray of numpy.array is used for graph data from this chapter.

Legend (loc, bbox_to_anchor)

The prepared array is ndarray of numpy.array, a 2-dimensional array with 4 rows and 2 columns.
The sum of the arrays is calculated for each row, and the result is a 1D array of 4 elements, which is used for the graph.

The legend is positioned in the center left using loc='center left' and bbox_to_anchor=(1, 0, 0.5, 1) to move the legend by one graph pixel to the right in the x direction.

A detailed explanation of the legend will be provided in another article.

import matplotlib.pyplot as plt
import numpy as np

# step1 Create label and data
labels = ['A', 'B', 'C', 'D']
width = 0.3
vals = np.array([[60, 32], [40, 35], [29, 10], [18, 5]])
# step2 Create graph frames
fig, ax = plt.subplots()
# step3 Plot a donut chart
ax.pie(vals.sum(axis=1), labels=labels, 
        wedgeprops={'width':width},
        startangle=90
)

ax.set_title('Donut chart with a legend')
# Legend
ax.legend(loc='center left', bbox_to_anchor=(1, 0, 0.5, 1))

plt.show()

Direction of rotation (counterclock)

The rotation direction of the pie chart can be clockwise by setting counterclock=False in the Axes.pie function.

# step3 Plot a donut chart
ax.pie(vals.sum(axis=1), labels=labels,
        wedgeprops={'width':width},
        startangle=90, 
        counterclock=False
)

Start position of circle (startangle)

The circumferential starting position of the element is specified as an angle in startangle.

The basic starting position is at (x, y) = (1, 0)

# step3 Plot a donut chart
ax.pie(vals.sum(axis=1), labels=labels,
        wedgeprops={'width':width},
        startangle=0, 
        counterclock=False
)

Numeric labels for each group (autopct, pctdistance)

Numeric labels can be displayed by specifying autopct as an argument to Axes.pie.

pctdistance can also change the position of the numerical labels.

# step3 Plot a donut chart
ax.pie(vals.sum(axis=1), labels=labels, startangle=90, counterclock=False,
        wedgeprops={'width':width},
        # Display of numeric labels
        autopct='%.1f%%', 
        # Position of numeric labels
        pctdistance=0.85,
)

Double donut graph (radius)

When data is divided into large and medium categories, the doughnut graph is doubled so that each element can be viewed in detail.

In Matplotlib, the inner donut graph is drawn by a combination of reducing the radius + separating the colors in the color map

The colormap used is tab20c, with a significant color change for each of the four.

Outer donut chart

It is drawn in the same way as the donut charts in the previous chapters.

Since there are four sets of colors per colormap, the array is passed to the colormap in multiples of four to create a color array of just the main colors.
cmap(np.arange(vals.shape[0])*4)=cmap([0, 4, 8, 12])

# step1 Create labels and data
labels = ['A', 'B', 'C', 'D']
width = 0.3
vals = np.array([[60, 32], [40, 35], [29, 10], [18, 5]])
# step2 Create graph frames
fig, ax = plt.subplots()
# step3 Create a colormap
cmap = plt.colormaps['tab20c']
# Color of an outer donut chart
outer_colors = cmap(np.arange(vals.shape[0])*4)
# vals.shape[0]=nrows
# np.arange(4)=[0, 1, 2, 3]
# cmap([0, 4, 8, 12])

# step4 Plot an outer donut chart
ax.pie(vals.sum(axis=1), radius=1, startangle=90, counterclock=False,
    labels=labels,
    autopct='%.0f%%', 
    pctdistance=1.15-width, 
    wedgeprops={'width':width, 'edgecolor':'white'},
    colors=outer_colors)

plt.show()

Inner donut chart (radius)

Since the main color is a multiple of 4, assign a multiple of 4 to each row and +1 to each column.
cmap([1, 2, 5, 6, 9, 10, 13, 14])

Data is made into one-dimensional array by numpy.flatten

radius=1-width reduces the radius of the graph by the width of the outer graph

# step3 Create a colormap
cmap = plt.colormaps['tab20c']
# Color an inner donut chart
inner_colors = cmap([i*4+j+1 for i, vs in enumerate(vals) for j in range(vs.size)])
# i*4: Assign multiples of 4 to each row
# j+1: Add +1 per column

# step5 Plot an inner donut chart
ax.pie(vals.flatten(), radius=1-width, startangle=90, counterclock=False,
    autopct='%.0f%%', 
    pctdistance=1.1-width, 
    wedgeprops={'width':width, 'edgecolor':'white'},
    colors=inner_colors)

double donut graph

A double donut graph plots the outer doughnut graph and the inner doughnut graph together.

The following tabs explain the code and flowchart.

• Python code
• Flowchart

# step1 Create labels and data
labels = ['A', 'B', 'C', 'D']
width = 0.3
vals = np.array([[60, 32], [40, 35], [29, 10], [18, 5]])
# step2 Create graph frames
fig, ax = plt.subplots()
# step3 Create a colormap
cmap = plt.colormaps['tab20c']
outer_colors = cmap(np.arange(vals.shape[0])*4)
inner_colors = cmap([i*4+j+1 for i, vs in enumerate(vals) for j in range(vs.size)])

# step4 Plot an outer donut chart
ax.pie(vals.sum(axis=1), radius=1, startangle=90, counterclock=False,
    labels=labels,
    autopct='%.0f%%', 
    pctdistance=1.15-width, 
    wedgeprops={'width':width, 'edgecolor':'white'},
    colors=outer_colors)

# step5 Plot an inner donut chart
ax.pie(vals.flatten(), radius=1-width, startangle=90, counterclock=False,
    autopct='%.0f%%', 
    pctdistance=1.1-width, 
    wedgeprops={'width':width, 'edgecolor':'white'},
    colors=inner_colors)

ax.set_title('tab20c in double donut chart')
ax.legend(loc='center left', bbox_to_anchor=(1, 0, 0.5, 1))
plt.show()

STEP

Create a colomap

Since it is hard to specify colors one by one, we create an array of colors by specifying numerical values in a colormap

For the color map, I used the following tab20c, which has four sets of one color

outer_colors

Since the colormap has four sets of one color, passing an array to the colormap in multiples of 4 will produce a color array with only the main color

cmap([0, 4, 8, 12])

inner_colors

Since the main color is a multiple of 4, assign a multiple of 4 to each row and +1 to each column

cmap([1, 2, 5, 6, 9, 10, 13, 14])

STEP

Plot an outer donut chart

The data in the 2D array of step1 are added by numpy.sum to make a 1D array and display a donut chart with radius 1.

The following are specified as arguments

x=vals.sum(axis=1)
Data : 1D array of [92, 75, 39, 23]

radius=1
Graph Radius

pctdistance=1.15-width
Numeric Label Distance

colors=outer_colors
Color of each element

STEP

Plot an inner donut chart

The data in the 2D array of step1 is turned into a 1D array by numpy.flatten, and a donut graph with radius=1-width is displayed.

The following are specified as arguments

x=vals.flatten()
Data: 1D array of [60, 32, 40, 35, 29, 10, 18, 5]

radius=1-width
Radius of circle (subtract by outer graph width)

pctdistance=1.1-width
Numeric Label Distance

colors=inner_colors
Color of each element

References

Donut pie chart

Labeled pie charts and donut pie charts