Assign specific colours to data in Matplotlib pie chart

Question

I'm trying to create pie charts with matplotlib in which the colour of each category is fixed.

I've got a function which creates a pie chart from sets of value and category data. Here's one example:

Category     Value
TI           65
Con          43
FR           40
TraI         40
Bug          38
Data         22
Int          15
KB           12
Other        8
Dep          7
PW           6
Uns          5
Perf         4
Dep          3

The problem is that the data differs from one instance to another, and that in turn changes the order of the categories. Thus, each category gets labelled a different colour each time I generate a chart. I could sort the data alphabetically every time, but that causes two problems: some categories are missing from some datasets, and I'd prefer it sorted by size anyway so that the smallest wedges are oriented horizontally.

How can I set matplotlib to assign colours depending on, say, the index of a pandas.Series?

Here's the code that I'm using to generate a pie chart:

import matplotlib.pyplot as plt

slices = [62, 39, 39, 38, 37, 21, 15,  9,  6,  7,  6,  5,  4, 3]

cmap = plt.cm.prism
colors = cmap(np.linspace(0., 1., len(slices)))

labels = [u'TI', u'Con', u'FR', u'TraI', u'Bug', u'Data', u'Int', u'KB', u'Other', u'Dep', u'PW', u'Uns', u'Perf', u'Dep']

fig = plt.figure(figsize=[10, 10])
ax = fig.add_subplot(111)

pie_wedge_collection = ax.pie(slices, colors=colors, labels=labels, labeldistance=1.05, autopct=make_autopct(slices))

for pie_wedge in pie_wedge_collection[0]:
    pie_wedge.set_edgecolor('white')

titlestring = 'Issues'

ax.set_title(titlestring)

EDIT: I forgot to explain the autopct function, it's for adding value and percentage labels:

def make_autopct(values):
    def my_autopct(pct):
        total = sum(values)
        val = int(round(pct*total/100.0))
        return '{p:.2f}%  ({v:d})'.format(p=pct,v=val)
    return my_autopct

Solution 1

Here's an idea you could try. Make a dictionary from your labels and colors, so each color is mapped to a label. Then, after making the pie chart, go in an assign the facecolor of the wedge using this dictionary.

Here's an untested bit of code which might do what you are looking for:

import numpy as np
import matplotlib.pyplot as plt

def mypie(slices,labels,colors):

    colordict={}
    for l,c in zip(labels,colors):
        print l,c
        colordict[l]=c

    fig = plt.figure(figsize=[10, 10])
    ax = fig.add_subplot(111)

    pie_wedge_collection = ax.pie(slices, labels=labels, labeldistance=1.05)#, autopct=make_autopct(slices))

    for pie_wedge in pie_wedge_collection[0]:
        pie_wedge.set_edgecolor('white')
        pie_wedge.set_facecolor(colordict[pie_wedge.get_label()])

    titlestring = 'Issues'

    ax.set_title(titlestring)

    return fig,ax,pie_wedge_collection

slices = [37, 39, 39, 38, 62, 21, 15,  9,  6,  7,  6,  5,  4, 3]
cmap = plt.cm.prism
colors = cmap(np.linspace(0., 1., len(slices)))
labels = [u'TI', u'Con', u'FR', u'TraI', u'Bug', u'Data', u'Int', u'KB', u'Other', u'Dep', u'PW', u'Uns', u'Perf', u'Dep']

fig,ax,pie_wedge_collection = mypie(slices,labels,colors)

plt.show()

Solution 2

Here is a simpler solution to @tmdavison's answer.

Let's first see the problem with an MWE:

import matplotlib.pyplot as plt

labels = ['Frogs', 'Hogs', 'Dogs', 'Logs']
sizes = [15, 30, 45, 10]

fig, ax = plt.subplots(1, 2)

ax[0].pie(sizes, labels=labels)
ax[1].pie(sizes[1:], labels=labels[1:])

This produces the problem plots:

The problem is that in the left-hand plot, Hogs is coloured in orange, but in the right-hand plot Hogs is coloured in blue (with a similar mix-up for Logs and Dogs).

We would like the colours for the labels to be the same across both plots. We can do this by specifying a dictionary of colours to use:

labels = ['Frogs', 'Hogs', 'Dogs', 'Logs']
sizes = [15, 30, 45, 10]
colours = {'Frogs': 'C0',
           'Hogs': 'C1',
           'Dogs': 'C2',
           'Logs': 'C3'}

fig, ax = plt.subplots(1, 2)

ax[0].pie(sizes,
          labels=labels,
          colors=[colours[key] for key in labels])

ax[1].pie(sizes[1:],
          labels=labels[1:],
          colors=[colours[key] for key in labels[1:]])

This works to create the plot:

Here we see that the labels are represented by the same colours across both plots, as desired.

If you have lots of categories it can be cumbersome to manually set a colour for each category. In this case you could construct the colours dictionary as:

colours = dict(zip(labels, plt.cm.tab10.colors[:len(labels)]))

If you have more than 10 categories you would instead use:

colours = dict(zip(labels, plt.cm.tab20.colors[:len(labels)]))