I am trying to make a discrete colorbar for a scatterplot in matplotlib

I have my x, y data and for each point an integer tag value which I want to be represented with a unique colour, e.g.

plt.scatter(x, y, c=tag)

typically tag will be an integer ranging from 0-20, but the exact range may change

so far I have just used the default settings, e.g.

plt.colorbar()

which gives a continuous range of colours. Ideally i would like a set of n discrete colours (n=20 in this example). Even better would be to get a tag value of 0 to produce a gray colour and 1-20 be colourful.

I have found some ‘cookbook’ scripts but they are very complicated and I cannot think they are the right way to solve a seemingly simple problem

You can create a custom discrete colorbar quite easily by using a BoundaryNorm as normalizer for your scatter. The quirky bit (in my method) is making 0 showup as grey.

For images i often use the cmap.set_bad() and convert my data to a numpy masked array. That would be much easier to make 0 grey, but i couldnt get this to work with the scatter or the custom cmap.

As an alternative you can make your own cmap from scratch, or read-out an existing one and override just some specific entries.

import numpy as np
import matplotlib as mpl
import matplotlib.pylab as plt

fig, ax = plt.subplots(1, 1, figsize=(6, 6))  # setup the plot

x = np.random.rand(20)  # define the data
y = np.random.rand(20)  # define the data
tag = np.random.randint(0, 20, 20)
tag[10:12] = 0  # make sure there are some 0 values to show up as grey

cmap = plt.cm.jet  # define the colormap
# extract all colors from the .jet map
cmaplist = [cmap(i) for i in range(cmap.N)]
# force the first color entry to be grey
cmaplist[0] = (.5, .5, .5, 1.0)

# create the new map
cmap = mpl.colors.LinearSegmentedColormap.from_list(
    'Custom cmap', cmaplist, cmap.N)

# define the bins and normalize
bounds = np.linspace(0, 20, 21)
norm = mpl.colors.BoundaryNorm(bounds, cmap.N)

# make the scatter
scat = ax.scatter(x, y, c=tag, s=np.random.randint(100, 500, 20),
                  cmap=cmap, norm=norm)

# create a second axes for the colorbar
ax2 = fig.add_axes([0.95, 0.1, 0.03, 0.8])
cb = plt.colorbar.ColorbarBase(ax2, cmap=cmap, norm=norm,
    spacing='proportional', ticks=bounds, boundaries=bounds, format='%1i')

ax.set_title('Well defined discrete colors')
ax2.set_ylabel('Very custom cbar [-]', size=12)

I personally think that with 20 different colors its a bit hard to read the specific value, but thats up to you of course.

You could follow this example:

#!/usr/bin/env python
"""
Use a pcolor or imshow with a custom colormap to make a contour plot.

Since this example was initially written, a proper contour routine was
added to matplotlib - see contour_demo.py and
http://matplotlib.sf.net/matplotlib.pylab.html#-contour.
"""

from pylab import *


delta = 0.01
x = arange(-3.0, 3.0, delta)
y = arange(-3.0, 3.0, delta)
X,Y = meshgrid(x, y)
Z1 = bivariate_normal(X, Y, 1.0, 1.0, 0.0, 0.0)
Z2 = bivariate_normal(X, Y, 1.5, 0.5, 1, 1)
Z = Z2 - Z1 # difference of Gaussians

cmap = cm.get_cmap('PiYG', 11)    # 11 discrete colors

im = imshow(Z, cmap=cmap, interpolation='bilinear',
            vmax=abs(Z).max(), vmin=-abs(Z).max())
axis('off')
colorbar()

show()

which produces the following image:

The above answers are good, except they don’t have proper tick placement on the colorbar. I like having the ticks in the middle of the color so that the number -> color mapping is more clear. You can solve this problem by changing the limits of the matshow call:

import matplotlib.pyplot as plt
import numpy as np

def discrete_matshow(data):
    #get discrete colormap
    cmap = plt.get_cmap('RdBu', np.max(data)-np.min(data)+1)
    # set limits .5 outside true range
    mat = plt.matshow(data,cmap=cmap,vmin = np.min(data)-.5, vmax = np.max(data)+.5)
    #tell the colorbar to tick at integers
    cax = plt.colorbar(mat, ticks=np.arange(np.min(data),np.max(data)+1))

#generate data
a=np.random.randint(1, 9, size=(10, 10))
discrete_matshow(a)

To set a values above or below the range of the colormap, you’ll want to use the set_over and set_under methods of the colormap. If you want to flag a particular value, mask it (i.e. create a masked array), and use the set_bad method. (Have a look at the documentation for the base colormap class: http://matplotlib.org/api/colors_api.html#matplotlib.colors.Colormap )

It sounds like you want something like this:

import matplotlib.pyplot as plt
import numpy as np

# Generate some data
x, y, z = np.random.random((3, 30))
z = z * 20 + 0.1

# Set some values in z to 0...
z[:5] = 0

cmap = plt.get_cmap('jet', 20)
cmap.set_under('gray')

fig, ax = plt.subplots()
cax = ax.scatter(x, y, c=z, s=100, cmap=cmap, vmin=0.1, vmax=z.max())
fig.colorbar(cax, extend='min')

plt.show()

This topic is well covered already but I wanted to add something more specific : I wanted to be sure that a certain value would be mapped to that color (not to any color).

It is not complicated but as it took me some time, it might help others not lossing as much time as I did :)

import matplotlib
from matplotlib.colors import ListedColormap

# Let's design a dummy land use field
A = np.reshape([7,2,13,7,2,2], (2,3))
vals = np.unique(A)

# Let's also design our color mapping: 1s should be plotted in blue, 2s in red, etc...
col_dict={1:"blue",
          2:"red",
          13:"orange",
          7:"green"}

# We create a colormar from our list of colors
cm = ListedColormap([col_dict[x] for x in col_dict.keys()])

# Let's also define the description of each category : 1 (blue) is Sea; 2 (red) is burnt, etc... Order should be respected here ! Or using another dict maybe could help.
labels = np.array(["Sea","City","Sand","Forest"])
len_lab = len(labels)

# prepare normalizer
## Prepare bins for the normalizer
norm_bins = np.sort([*col_dict.keys()]) + 0.5
norm_bins = np.insert(norm_bins, 0, np.min(norm_bins) - 1.0)
print(norm_bins)
## Make normalizer and formatter
norm = matplotlib.colors.BoundaryNorm(norm_bins, len_lab, clip=True)
fmt = matplotlib.ticker.FuncFormatter(lambda x, pos: labels[norm(x)])

# Plot our figure
fig,ax = plt.subplots()
im = ax.imshow(A, cmap=cm, norm=norm)

diff = norm_bins[1:] - norm_bins[:-1]
tickz = norm_bins[:-1] + diff / 2
cb = fig.colorbar(im, format=fmt, ticks=tickz)
fig.savefig("example_landuse.png")
plt.show()

I have been investigating these ideas and here is my five cents worth. It avoids calling BoundaryNorm as well as specifying norm as an argument to scatter and colorbar. However I have found no way of eliminating the rather long-winded call to matplotlib.colors.LinearSegmentedColormap.from_list.

Some background is that matplotlib provides so-called qualitative colormaps, intended to use with discrete data. Set1, e.g., has 9 easily distinguishable colors, and tab20 could be used for 20 colors. With these maps it could be natural to use their first n colors to color scatter plots with n categories, as the following example does. The example also produces a colorbar with n discrete colors approprately labelled.

import matplotlib, numpy as np, matplotlib.pyplot as plt
n = 5
from_list = matplotlib.colors.LinearSegmentedColormap.from_list
cm = from_list(None, plt.cm.Set1(range(0,n)), n)
x = np.arange(99)
y = x % 11
z = x % n
plt.scatter(x, y, c=z, cmap=cm)
plt.clim(-0.5, n-0.5)
cb = plt.colorbar(ticks=range(0,n), label='Group')
cb.ax.tick_params(length=0)

which produces the image below. The n in the call to Set1 specifies the first n colors of that colormap, and the last n in the call to from_list specifies to construct a map with n colors (the default being 256). In order to set cm as the default colormap with plt.set_cmap, I found it to be necessary to give it a name and register it, viz:

cm = from_list('Set15', plt.cm.Set1(range(0,n)), n)
plt.cm.register_cmap(None, cm)
plt.set_cmap(cm)
...
plt.scatter(x, y, c=z)

I think you’d want to look at colors.ListedColormap to generate your colormap, or if you just need a static colormap I’ve been working on an app that might help.