CODEWITHSUNDEEP
pythonnumpymatplotlibplot
Lee
Lee

Reputation: 31050

Customising matplotlib cmaps

I have some normalised histogram data in array of shape (12,1):

>>> hnorm

   array([[ 0.        ],
       [ 0.        ],
       [ 0.01183432],
       [ 0.0295858 ],
       [ 0.04142012],
       [ 0.04142012],
       [ 0.03550296],
       [ 0.01775148],
       [ 1.        ],
       [ 0.98816568],
       [ 0.56213018],
       [ 0.        ]])

I'd like to plot this in 'heatmap' style. I am doing this like so:

import matplotlib.pyplot as plt
plt.imshow(hnorm, cmap='RdBu',origin='lower')

This works (axis formatting aside).

enter image description here

However, I'd like to customise the colormap to fade from white to Red. I have attempted:

import matplotlib.colors as col

cdict = {'red': ((0.00, 0.07, 0.14),
        (0.21, 0.28, 0.35),
        (0.42, 0.49, 0.56),
        (0.63, 0.70, 0.77),
        (0.84, 0.91, 0.99)),
        'green': ((0.0, 0.0, 0.0),
        (0.0, 0.0, 0.0),
        (0.0, 0.0, 0.0),
        (0.0, 0.0, 0.0),
        (0.0, 0.0, 0.0)),
        'blue': ((0.0, 0.0, 0.0),
        (0.0, 0.0, 0.0),
        (0.0, 0.0, 0.0),
        (0.0, 0.0, 0.0), 
        (0.0, 0.0, 0.0))}
cmap1 = col.LinearSegmentedColormap('my_colormap',cdict,N=256,gamma=0.75)
plt.imshow(hnorm, cmap=cmap1,origin='lower')

This fails. Any ideas what I am doing wrong?

Upvotes: 1

Views: 1638

Answers (1)

Rutger Kassies
Rutger Kassies

Reputation: 64443

The cmap 'Reds' as askewchan is suggesting is simpler and (imo) also better looking. But i'll answer just to show how your approach of building a custom cmap could also work.

In your color dict you have 5 entries at which you specify the color. Since you want to only use red and white you need only two enties. For white, all colors must be used which is specified by color values of 1.0 at position 0.0. For red only red should be used at position 1.0.

You also only provide values (other than 0) for your red tuple. This will only give you different shades of red between 'full' red and black (since you always have green and blue as 0).

cdict = {'red': ((0.0, 1.0, 1.0),
                 (1.0, 1.0, 1.0)),

        'green': ((0.0, 1.0, 1.0),
                  (1.0, 0.0, 0.0)),

        'blue': ((0.0, 1.0, 1.0),
                 (1.0, 0.0, 0.0))}

my_cmap = mpl.colors.LinearSegmentedColormap('my_colormap', cdict)

plt.imshow(np.random.rand(20,20), cmap=my_cmap, origin='lower', interpolation='none')
plt.colorbar(shrink=.75)

enter image description here

Another example showing how the two color items allow 'jumps' in the cmap:

cdict = {'red': ((0.0, 1.0, 1.0), # full red
                 (0.5, 1.0, 0.0), # full red till, no red after
                 (1.0, 1.0, 1.0)), # full red

        'green': ((0.0, 1.0, 1.0), # full green
                  (0.5, 0.0, 0.0), # no green
                  (1.0, 1.0, 1.0)), # full green

        'blue': ((0.0, 1.0, 1.0), # full blue
                 (0.5, 0.0, 1.0), # no blue till, full blue after
                 (1.0, 1.0, 1.0))} # full blue

my_cmap = mpl.colors.LinearSegmentedColormap('my_colormap', cdict)

plt.imshow(np.random.rand(20,20), cmap=my_cmap, origin='lower', interpolation='none')
plt.colorbar(shrink=.75)

enter image description here

Upvotes: 5

Related Questions