Plot a Basic 2D Histogram using Matplotlib

2D density/histogram are charts used to display relationship between 2 numerical variables when there are lots of data points. Scatter plots cannot really be used in this case due to overplotting in the chart.

This post is dedicated to 2D histograms made with matplotlib, through the hist2D() function. You'll learn how to customize bin sizes, control colors and add a legend.

Libraries & Dataset

Let's start by import a few libraries and create a dataset:

# libraries
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
 
# create data
size = 100000
df = pd.DataFrame({
   'x': np.random.normal(size=size),
   'y': np.random.normal(size=size)
})
df.head()

	x	y
0	0.156635	0.497530
1	-0.485384	-1.329300
2	-1.116573	1.873535
3	0.841880	0.375499
4	-0.528407	-1.696453

2D histograms

2D histograms are useful when you need to analyse the relationship between 2 numerical variables that have a huge number of values. It is useful for avoiding the over-plotted scatterplots.

The following example illustrates the importance of the bins argument. You can explicitly tell how many bins you want for the X and the Y axis.

The parameters of hist2d() function used in the example are:

x, y: input values
bins: the number of bins in each dimension
cmap : colormap

fig, axs = plt.subplots(nrows=2, ncols=2, figsize=(8,8))

# Big bins
axs[0,0].hist2d(x, y, bins=(50, 50), cmap=plt.cm.jet)
axs[0, 0].set_title('bins = (50, 50)')
 
# Small bins
axs[0,1].hist2d(x, y, bins=(600, 600), cmap=plt.cm.jet)
axs[0, 1].set_title('bins = (600, 600)')
 
# If you do not set the same values for X and Y, the bins won't be a square!
axs[1,0].hist2d(x, y, bins=(600, 30), cmap=plt.cm.jet)
axs[1, 0].set_title('bins = (600, 30)')

# If you do not set the same values for X and Y, the bins won't be a square!
axs[1,1].hist2d(x, y, bins=(30, 600), cmap=plt.cm.jet)
axs[1, 1].set_title('bins = (30, 600)')

plt.show()

Colors

Once you decide the bin size, it is possible to change the colour palette. Matplolib provides a whole bunch of pre-defined color map (also know as cmap).

Here you can find how to use them in a 2d histogram:

fig, axs = plt.subplots(nrows=2, ncols=2, figsize=(8,8))

# Big bins
axs[0,0].hist2d(x, y, bins=(50, 50), cmap=plt.cm.Reds_r)
axs[0, 0].set_title('cmap=plt.cm.Reds')
 
# Small bins
axs[0,1].hist2d(x, y, bins=(50, 50), cmap=plt.cm.Blues_r)
axs[0, 1].set_title('cmap=plt.cm.Blues')
 
# If you do not set the same values for X and Y, the bins won't be a square!
axs[1,0].hist2d(x, y, bins=(50, 50), cmap=plt.cm.Greens_r)
axs[1, 0].set_title('cmap=plt.cm.Greens')

# If you do not set the same values for X and Y, the bins won't be a square!
axs[1,1].hist2d(x, y, bins=(50, 50), cmap=plt.cm.Greys_r)
axs[1, 1].set_title('cmap=plt.cm.Greys')

plt.show()

Colorbar

Finally, it might be useful to add a color bar on the side as a legend. You can add a color bar using colorbar() function.

plt.hist2d(x, y, bins=(50, 50), cmap=plt.cm.Greys_r)
plt.colorbar()
plt.show()

Going further

🚨 Grab the Data To Viz poster!

Do you know all the chart types? Do you know which one you should pick? I made a decision tree that answers those questions. You can download it for free!