Python Pandas - Scatter Matrix

---

---

A scatter matrix, also known as a pair plot, is a grid of scatter plots that helps in visualizing relationships between multiple variables at once. Each cell in the matrix represents a scatter plot of two variables, and diagonal plots often represent distributions, like histograms. This plot is useful tool for visualizing the correlation between features in a dataset.

In this tutorial, we will learn about what is the Pandas method for creating scatter matrix and how to customize the scatter matrix according to different needs.

Scatter Matrix in Pandas

Pandas provides a simple method to generate a scatter matrix of a DataFrame using the pandas.plotting.scatter_matrix() function. And it results an NumPy array np.ndarray of scatter plots representing pairwise relationships between columns.

Syntax

The basic syntax for creating a scatter matrix is as follows −

pandas.plotting.scatter_matrix(frame, alpha=0.5, figsize=None, ax=None, grid=False, diagonal='hist', marker='.', density_kwds=None, hist_kwds=None, range_padding=0.05, **kwargs)

Where,

• frame: DataFrame, the input dataset that you want to visualize.

• alpha: An optional parameter used to specify the amount of transparency applied to the scatter plot (between 0 and 1).

• figsize: An optional parameter determines the dimensions of the figure in inches (width, height).

• ax: An optional Matplotlib axis object, if provided, the scatter matrix will be drawn on this axis.

• grid: If the boolean value True is specifies, then it displays the grid in the scatter matrix.

• diagonal: Determines whether to show histograms ('hist') or Kernel Density Estimation plots ('kde') on the diagonal.

• marker: Specifies the marker style for the scatter plots (default is '.').

• density_kwds: Additional keyword arguments for the KDE plot.

• hist_kwds: Additional keyword arguments for the histogram plot.

• range_padding: Relative extension of axis range (default is 0.05).

• **kwargs: Additional arguments to customize the plot.

Example

In this example, we will create a scatter matrix using the pandas.plotting.scatter_matrix() function with a sample dataset.

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

plt.rcParams["figure.figsize"] = [7, 4]

# Sample dataset creation
np.random.seed(0)

#create DataFrame
df = pd.DataFrame({'points': np.random.randn(1000),
                   'assists': np.random.randn(1000),
                   'rebounds': np.random.randn(1000)})

# Create a scatter matrix
pd.plotting.scatter_matrix(df, alpha=0.5, diagonal='hist', grid=True)

# Display the plot
plt.show()

After executing the above code, we get the following output −

Customizing the Scatter Matrix

Pandas allows customization of the scatter matrix by adjusting various parameters of the pandas.plotting.satter_matrix() function.

Example: Customize Scatter Matrix Marker Style and Transparency

The following example demonstrates how to customize the scatter matrix marker style and transparency.

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

plt.rcParams["figure.figsize"] = [7, 4]

# Sample dataset creation
np.random.seed(0)

#create DataFrame
df = pd.DataFrame({'points': np.random.randn(1000),
                   'assists': np.random.randn(1000),
                   'rebounds': np.random.randn(1000)})

# Create a scatter matrix
pd.plotting.scatter_matrix(df, alpha=0.7, diagonal='hist', marker='D', grid=True)

# Display the plot
plt.show()

Following is the output of the above code −

Example: Customizing Scatter Matrix Color

The following example demonstrates customizing the scatter matrix color by using the color parameter of the plotting.scatter_matrix() function.

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

plt.rcParams["figure.figsize"] = [7, 4]

# Sample dataset creation
np.random.seed(0)

#create DataFrame
df = pd.DataFrame({'points': np.random.randn(1000),
                   'assists': np.random.randn(1000),
                   'rebounds': np.random.randn(1000)})

# Create a scatter matrix
pd.plotting.scatter_matrix(df, alpha=0.7, diagonal='hist', color='green', grid=True)

# Display the plot
plt.show()

Following is the output of the above code −

Change the Diagonal Plot of the Scatter Matrix

You can change the default histogram plot to the Kernel Density Estimation (kde) using the diagonal parameter of the plotting.scatter_matrix() function.

Example

The following example demonstrates replacing the histograms on the diagonal with Kernel Density Estimation (kde) plots.

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

plt.rcParams["figure.figsize"] = [7, 4]

# Sample dataset creation
np.random.seed(0)

#create DataFrame
df = pd.DataFrame({'points': np.random.randn(1000),
                   'assists': np.random.randn(1000),
                   'rebounds': np.random.randn(1000)})

# Create a scatter matrix
pd.plotting.scatter_matrix(df, alpha=0.7, diagonal='kde', grid=True)

# Display the plot
plt.show()

Following is the output of the above code −