Pairplot in Matplotlib

Pair Plot is a type of chart that shows how different numbers in a dataset relate to each other. It creates multiple small scatter plots, comparing two variables at a time. While Seaborn has a ready-made pairplot() function to quickly create this chart, Matplotlib allows more control to customize how the plot looks and behaves. A Pair Plot (also called a scatterplot matrix) consists of:

• Scatter plots for each pair of numerical variables.
• Histograms (or kernel density plots) on the diagonal, representing the distribution of individual variables.

This visualization helps in identifying:

• Linear and non-linear relationships between features.
• Clusters or groups within data.
• Potential outliers.

Creating a pair plot using matplotlib

To get started, we first need to import the necessary libraries.

import matplotlib.pyplot as plt

import pandas as pd

import numpy as np

• matplotlib.pyplot: Used for creating visualizations.
• pandas: Helps in handling structured data (dataframes).
• numpy: Useful for generating numerical data.

Implementation:

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

np.random.seed(42)
data = pd.DataFrame({
    'Feature 1': np.random.rand(50),
    'Feature 2': np.random.rand(50),
    'Feature 3': np.random.rand(50),
    'Feature 4': np.random.rand(50)
})

# Number of features
num_features = len(data.columns)

# Create Subplots Grid
fig, axes = plt.subplots(num_features, num_features, figsize=(10, 10))

# Loop through each pair of features
for i in range(num_features):
    for j in range(num_features):
        ax = axes[i, j]
        
        if i == j:
            # Diagonal: Histogram of the feature
            ax.hist(data.iloc[:, i], bins=15, color='skyblue', edgecolor='black')
        else:
            # Scatter plot for feature pairs
            ax.scatter(data.iloc[:, j], data.iloc[:, i], alpha=0.7, s=10, color="blue")

        # Set labels on the left and bottom axes
        if j == 0:
            ax.set_ylabel(data.columns[i], fontsize=10)
        if i == num_features - 1:
            ax.set_xlabel(data.columns[j], fontsize=10)

        # Remove ticks for a cleaner look
        ax.set_xticks([])
        ax.set_yticks([])

# Adjust layout
plt.tight_layout()
plt.show()

Output

Explanation:

• Data Generation: 4 features × 50 values (0-1) stored in a Pandas DataFrame (np.random.seed(42)).
• Subplots Grid: 4×4 layout (plt.subplots()), with histograms on the diagonal (i == j) and scatter plots elsewhere (i ≠ j).
• Histograms: ax.hist() with 15 bins, skyblue fill, black edges for clarity.
• Scatter Plots: ax.scatter() with alpha=0.7, s=10, blue color to show relationships.
• Formatting: Labels only on leftmost column (j == 0) & bottom row (i == num_features - 1). Ticks removed for a clean look. plt.tight_layout() prevents overlap.
• plt.show() renders the final visualization.

Advantages of pair plot in matplotlib

• Customizability: Unlike Seaborn’s pairplot(), Matplotlib allows full control over plot styling.
• Better Integration: Works seamlessly within larger Matplotlib-based visualizations.
• Flexibility: Can modify elements like colors, markers, line styles, and annotations easily.

Enhancing the pair plot

To improve the visualization, consider:

• Adding regression lines to scatter plots.
• Using different colors to highlight categories in the dataset.
• Replacing histograms with kernel density estimation (KDE) plots.

Example:

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

np.random.seed(42)
data = pd.DataFrame(np.random.rand(50, 4), columns=['Feature 1', 'Feature 2', 'Feature 3', 'Feature 4'])

# Number of features
num_features = len(data.columns)

# Create figure
fig, axes = plt.subplots(num_features, num_features, figsize=(10, 10))

# Loop through each pair of features
for i in range(num_features):
    for j in range(num_features):
        ax = axes[i, j]

        if i == j:
            # Plot histogram on the diagonal
            ax.hist(data.iloc[:, i], bins=10, color="skyblue", edgecolor="black")
        else:
            # Scatter plot
            x = data.iloc[:, j]
            y = data.iloc[:, i]
            ax.scatter(x, y, alpha=0.7, s=10, color="blue")

            # Add Regression Line
            m, b = np.polyfit(x, y, 1)  # Linear regression
            ax.plot(x, m*x + b, color="red", linewidth=1)

        # Labels
        if j == 0:
            ax.set_ylabel(data.columns[i], fontsize=10)
        if i == num_features - 1:
            ax.set_xlabel(data.columns[j], fontsize=10)

        # Hide ticks for cleaner look
        ax.set_xticks([])
        ax.set_yticks([])

# Adjust layout
plt.tight_layout()
plt.show()

Output:

Explanation:

• Data Preparation: Random values are generated for four features using NumPy and Pandas DataFrame stores the dataset.
• Creating Subplots: A 4×4 grid of subplots is created to display the pairwise relationships. plt.subplots(num_features, num_features, figsize=(10, 10)) sets up the grid layout.
• Plotting the Pair Plot: If i == j, a histogram is plotted on the diagonal using ax.hist(). If i ≠ j, a scatter plot is created using ax.scatter().
• Adding Regression Lines: The np.polyfit(x, y, 1) function computes the slope (m) and intercept (b) of the regression line. The ax.plot(x, m*x + b, color="red", linewidth=1) function overlays a red regression line on the scatter plot.
• Labels are added to only the leftmost and bottom plots. Ticks are hidden for a clean design.
• plt.tight_layout() ensures proper spacing for readability.