How to create a 2D density plot color coded with the fraction of points that are in two dataframes

I have two 2D dataframes, defined in the (x,y) plane, df1 and df2 and they are located in a similar region of the x,y plane.

I would like to create a plot color coded with number of points in df2 divided by the number of points in df1.

I try something, but it looks pixelated, do you have any recommendation?

df1 = pd.DataFrame(np.random.random((10000,2)), columns=list('AB'))
df2 = pd.DataFrame(np.random.random((10000,2)), columns=list('AB'))
df1


upper_lim_A = 0.8
lower_lim_A = 0.3

upper_lim_B = 0.7
lower_lim_B = 0.2

fontsize = 8
N_bins = 10
hist_threshold = 1 #Number of points in a bin to be considered populated


%matplotlib inline
#Creating the histograms
hist_all, *edges_all = np.histogram2d(df1['B'], df1['A'], bins=N_bins, range = ((lower_lim_B, upper_lim_B), (lower_lim_A,upper_lim_A)))
hist_vars, *edges_vars = np.histogram2d(df2['B'], df2['A'], bins=N_bins, range = ((lower_lim_B, upper_lim_B), (lower_lim_A,upper_lim_A)))

fig, ax1 = plt.subplots( figsize=(10.5,2.5*3))
ax1.tick_params(direction='out', length=6, width=2, colors='k', grid_alpha=1, labelsize=size_la)

#Finding the ratio df2/df1
ratio = hist_vars/hist_all
#Replace Nan with 0
ratio[np.isnan(ratio)] = 0

#Deep copy the ratio array
ratio_copy = np.copy(ratio)
#Set the ratio to nan if hist_all < hist_threshold
ratio[hist_all < hist_threshold*0.9] = np.nan

color_style = 'inferno'
vmax = 1.1

im = ax1.imshow(ratio, alpha=1,origin='lower',cmap=color_style,vmin=0, aspect='auto', extent=[lower_lim_A,upper_lim_A,lower_lim_B, upper_lim_B])

from mpl_toolkits.axes_grid1 import make_axes_locatable

ax0 = fig.add_subplot(111)
ax0.set_frame_on(False)
ax0.xaxis.set_ticks([])
ax0.yaxis.set_ticks([])

# create a divider for the colorbar
divider = make_axes_locatable(ax0)
cax = divider.append_axes("top", size="5%", pad=0.1)

# add the colorbar to the new axis
cbar = fig.colorbar(im, cax=cax, orientation='horizontal')
cbar.ax.tick_params(labelsize=size_la)
cbar.set_label(r'$\mathrm{N}_{\mathrm{df2}}/\mathrm{N}_{\mathrm{df1}}$', fontsize=25)
cax.xaxis.set_ticks_position('top')
cax.xaxis.set_label_position('top')
ax1.invert_yaxis()

I removed much code that was only related for showing the image. There is already a parameter for the number of bins, that you can use to make the result smoother.

Also, please note that I changed the way the ratio is calculated. I guess first / (first + second) is a little more meaningful?

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

df1 = pd.DataFrame(np.random.random((10000, 2)), columns=list('AB'))
df2 = pd.DataFrame(np.random.random((10000, 2)), columns=list('AB'))


upper_lim_A = 0.8
lower_lim_A = 0.3

upper_lim_B = 0.7
lower_lim_B = 0.2


# increase this for more bins
N_bins = 40

hist1, *edges_all = np.histogram2d(
    df1['B'], df1['A'],
    bins=N_bins,
    range=((lower_lim_B, upper_lim_B), (lower_lim_A,upper_lim_A))
)

hist2, *edges_vars = np.histogram2d(
    df2['B'], df2['A'],
    bins=N_bins,
    range=((lower_lim_B, upper_lim_B), (lower_lim_A,upper_lim_A))
)

# ratio is now share of total (not share of hist 1) + small epsilon to avoid div by 0
ratio = hist2/(hist1 + hist2 + 10**-6)


plt.imshow(ratio)
plt.colorbar()

This is an alternative to lager bin number. This uses open cv, since it is very convenient, but you can use any image processing library that allows you to blur images (or even scipy).

import cv2
blurred_ratio = cv2.blur(ratio, ksize=(5,5))  # choose any kernel size for gaussian blur
plt.imshow(blurred_ratio)