Plot 2-D Histogram in Python using Matplotlib

2D Histogram is used to analyze the relationship among two data variables which has wide range of values. A 2D histogram is very similar like 1D histogram. The class intervals of the data set are plotted on both x and y axis. Unlike 1D histogram, it drawn by including the total number of combinations of the values which occur in intervals of x and y, and marking the densities. It is useful when there is a large amount of data in a discrete distribution, and simplifies it by visualizing the points where the frequencies if variables are dense.

Creating a 2D Histogram

Matplotlib library provides an inbuilt function matplotlib.pyplot.hist2d() which is used to create 2D histogram.Below is the syntax of the function:

matplotlib.pyplot.hist2d(x, y, bins=(nx, ny), range=None, density=False, weights=None, cmin=None, cmax=None, cmap=value)

Here (x, y) specify the coordinates of the data variables, the length of the X data and Y variables should be same.The number of bins can be specified by the attribute bins=(nx, ny) where nx and ny is the number of bins to be used in the horizontal and vertical directions respectively.cmap=value is used to set the color scale.The range=None is an optional parameter used to set rectangular area in which data values are counted for plot.density=value is optional parameter accepting boolean values used to normalize histogram.
The code below code creates a simple 2D histogram using matplotlib.pyplot.hist2d() function having some random values of x and y:

# Import libraries
import numpy as np
import matplotlib.pyplot as plt
import random

# Creating dataset
n = 100
x = np.random.standard_normal(n)
y = 3.0 * x

fig = plt.subplots(figsize =(10, 7))
# Creating plot
plot.hist2d(x, y)
plot.title("Simple 2D Histogram")

# show plot
plot.show()

Output:

Customizing 2D Histogram

The matplotlib.pyplot.hist2d() function has a wide range of methods which we can use to customize and create the plot for better view and understanding.

# Import libraries
import numpy as np
import matplotlib.pyplot as plt
import random

# Creating dataset
x = np.random.normal(size = 500000)
y = x * 3 + 4 * np.random.normal(size = 500000)

fig = plt.subplots(figsize =(10, 7))
# Creating plot
plot.hist2d(x, y)
plot.title("Simple 2D Histogram")

# show plot
plot.show()

Output:

Some of the customization of the above graph are listed below:

Changing the bin scale:-

# Import libraries
import numpy as np
import matplotlib.pyplot as plt
import random

# Creating dataset
x = np.random.normal(size = 500000)
y = x * 3 + 4 * np.random.normal(size = 500000)

# Creating bins
x_min = np.min(x)
x_max = np.max(x)

y_min = np.min(y)
y_max = np.max(y)

x_bins = np.linspace(x_min, x_max, 50)
y_bins = np.linspace(y_min, y_max, 20)

fig, ax = plt.subplots(figsize =(10, 7))
# Creating plot
plt.hist2d(x, y, bins =[x_bins, y_bins])
plt.title("Changing the bin scale")

ax.set_xlabel('X-axis')
ax.set_ylabel('X-axis')

# show plot
plt.tight_layout()
plot.show()

Output:

Changing the color scale and adding color bar:-

# Import libraries
import numpy as np
import matplotlib.pyplot as plt
import random

# Creating dataset
x = np.random.normal(size = 500000)
y = x * 3 + 4 * np.random.normal(size = 500000)

# Creating bins
x_min = np.min(x)
x_max = np.max(x)

y_min = np.min(y)
y_max = np.max(y)

x_bins = np.linspace(x_min, x_max, 50)
y_bins = np.linspace(y_min, y_max, 20)

fig, ax = plt.subplots(figsize =(10, 7))
# Creating plot
plt.hist2d(x, y, bins =[x_bins, y_bins], cmap = plt.cm.nipy_spectral)
plt.title("Changing the color scale and adding color bar")

# Adding color bar
plt.colorbar()

ax.set_xlabel('X-axis')
ax.set_ylabel('X-axis')

# show plot
plt.tight_layout()
plot.show()

Output:

Filtering data:-

# Import libraries
import numpy as np
import matplotlib.pyplot as plt
import random

# Creating dataset
x = np.random.normal(size = 500000)
y = x * 3 + 4 * np.random.normal(size = 500000)

# Creating bins
x_min = np.min(x)
x_max = np.max(x)

y_min = np.min(y)
y_max = np.max(y)

x_bins = np.linspace(x_min, x_max, 50)
y_bins = np.linspace(y_min, y_max, 20)

# Creating data filter
data = np.c_[x, y]

for i in range(10000):
    x_idx = random.randint(0, 500000)
    data[x_idx, 0] = -9999

data = data[data[:, 0]!=-9999]

fig, ax = plt.subplots(figsize =(10, 7))
# Creating plot
plt.hist2d(data[:, 0], data[:, 1], bins =[x_bins, y_bins])
plt.title("Filtering data")

ax.set_xlabel('X-axis')
ax.set_ylabel('X-axis')

# show plot
plt.tight_layout()
plot.show()

Output:

Using matplotlib hexbin function:-

# Import libraries
import numpy as np
import matplotlib.pyplot as plt
import random

# Creating dataset
x = np.random.normal(size = 500000)
y = x * 3 + 4 * np.random.normal(size = 500000)

fig, ax = plt.subplots(figsize =(10, 7))
# Creating plot
plt.title("Using matplotlib hexbin function")
plt.hexbin(x, y, bins = 50)

ax.set_xlabel('X-axis')
ax.set_ylabel('Y-axis')

# show plot
plt.tight_layout()
plot.show()

Output: