Heat maps in python - with colors

Question

I try to make a program for plotting a heat map based on the clicks from the participants. There are two bodies with increase and decrease emotions.

I want to show the intensity of clicks on the left body with blue color (more intense blue = more number of clicks) and right body with red color.

The problem is that I need to show it in one body and also with the background image over which I see the heat map.

x=blue[:,1]
y=blue[:,2]
ax = plt.gca()

heatmap, xedges, yedges = np.histogram2d(x, y, bins=50)
extent = [xedges[0], xedges[-1], yedges[0], yedges[-1]]
plt.imshow(heatmap.T, extent=extent, origin='lower')
ax = plt.gca()
ax.invert_yaxis()

plt.show()

x1=red[:,1]
y1=red[:,2]
ax = plt.gca()

heatmap, xedges, yedges = np.histogram2d(x1, y1, bins=50)
extent = [xedges[0], xedges[-1], yedges[0], yedges[-1]]

plt.clf()
plt.imshow(heatmap.T, extent=extent, origin='lower')
ax = plt.gca()
ax.invert_yaxis()

plt.show()
plt.imshow(image)

imageFile = cbook.get_sample_data('C:\\Users\\Admin\\Desktop\\pythonpca\\result.png')
image = plt.imread(imageFile)
plt.plot(all_samples[0:240,0],all_samples[0:240,1], 'o', markersize=3, color='blue', alpha=0.5, label='increase')
plt.imshow(image)

By this I get a heat map for the left body clicks, a heat map for right body clicks and a picture of the left and right bodies. I want them all in the same picture, with blue and red hotspots. I am attaching the pictures I get with this.

2 bodies picture (I have plotted blue points on it, but I do not need the points):

The heat maps from the left and right bodies:

Please let me know if I should add more info.

Answer 1

You could use two different colors to represent left vs right-body clicks, and use the opacity (alpha channel) to represent the density of clicks within a given region. One way to do this would be to create a custom LinearSegmentedColormap that varies in the alpha channel:

import numpy as np
from matplotlib import pyplot as plt
from matplotlib import colors

# Some fake data
nbins = (50, 50)
ranges = ((-2, 2), (-2, 2))
lx, ly = np.random.randn(2, 10000) * 0.5 + 1
rx, ry = np.random.randn(2, 10000) * 0.75 - 1

# Compute 2D histograms
left_density = np.histogram2d(lx, ly, bins=nbins, range=ranges, normed=True)[0]
right_density = np.histogram2d(rx, ry, bins=nbins, range=ranges, normed=True)[0]

# Make some custom colormaps that vary in the alpha channel.
trans2blue = colors.LinearSegmentedColormap.from_list(
    name='Trans2Blue', colors=[(0., 0., 1., 0.), (0., 0., 1., 1.)])
trans2red = colors.LinearSegmentedColormap.from_list(
    name='Trans2Red', colors=[(1., 0., 0., 0.), (1., 0., 0., 1.)])

# `imshow` the histograms using the custom colormaps.
fig, ax = plt.subplots(1, 1)
left_im = ax.imshow(left_density, cmap=trans2blue)
right_im = ax.imshow(right_density, cmap=trans2red)

right_cb = fig.colorbar(right_im)
right_cb.set_label('Right click density')
left_cb = fig.colorbar(left_im)
left_cb.set_label('Left click density')

# Workaround for https://stackoverflow.com/q/15003353/1461210
left_cb.solids.set_edgecolor("face")
right_cb.solids.set_edgecolor("face")

fig.tight_layout()

Another way to achieve the same effect would be to construct two (rows, cols, 4) arrays of RGBA pixel values where the alpha channel contains your density values, and RGB contains your desired color, then imshow these on top of one another. Using a custom colormap has a few advantages though - the colormap range gets scaled automatically, and it's fairly easy to add colorbars.