5 Best Ways to Make a Scatter Plot for Clustering in Python

by

💡 Problem Formulation: When working with clustering in Python, visualizing the distribution and grouping of data points is crucial for understanding the underlying patterns and structure. A scatter plot is an ideal tool for this purpose. This article explores how to create a scatter plot for datasets post-clustering, where the input is a set of data points with their cluster labels, and the desired output is a visual representation distinguishing the clusters.

Method 1: Using Matplotlib

Matplotlib is a comprehensive library for creating static, animated, and interactive visualizations in Python. For scatter plots specifically, Matplotlib’s scatter() function is versatile and allows customization of color, scale, and marker style. It is particularly well-suited to plot large datasets and customize plots with labels, legends, and multiple layers of information.

Here’s an example:

import matplotlib.pyplot as plt

# Sample cluster data
x = [1, 2, 3, 4, 5]
y = [5, 4, 3, 2, 1]
cluster_labels = [0, 0, 1, 1, 1]

# Create scatter plot
plt.scatter(x, y, c=cluster_labels)
plt.xlabel('Feature 1')
plt.ylabel('Feature 2')
plt.title('Scatter Plot for Clustering')
plt.colorbar()
plt.show()

Output: A scatter plot with points colored according to their cluster label and a color bar indicating cluster identity.

The code above uses x and y lists to represent feature values, with cluster_labels denoting the assigned cluster. Matplotlib’s scatter method maps these points onto a 2D space and uses the cluster labels to color them distinctly, aided by a color bar for reference.

Method 2: Using Seaborn

Seaborn provides a high-level interface for drawing attractive and informative statistical graphics, including scatter plots. It has enhanced support for more complicated visualizations involving categorical data and integrates well with Pandas DataFrames. Its scatterplot() function is particularly useful when dealing with DataFrames and allows easy mapping of clusters to colors.

Here’s an example:

import seaborn as sns
import pandas as pd

# Sample cluster data in DataFrame
data = pd.DataFrame({
    'Feature 1': [1, 2, 3, 4, 5],
    'Feature 2': [5, 4, 3, 2, 1],
    'Cluster': [0, 0, 1, 1, 1]
})

# Create scatter plot
sns.scatterplot(data=data, x='Feature 1', y='Feature 2', hue='Cluster')
plt.show()

Output: A scatter plot using Seaborn with data points colored by their respective cluster labels.

The Seaborn snippet creates a scatter plot from a DataFrame. Columns for ‘Feature 1’ and ‘Feature 2’ map to the axes, while the ‘Cluster’ column defines the color. Seaborn’s easy integration with Pandas makes it convenient for plotting directly from DataFrames.

Method 3: Using Plotly

Plotly is a graphing library that enables interactive scatter plots that can be used in web browsers. It offers highly interactive charts that can be zoomed, panned, and filtered. Plotly is powerful for creating dashboards and web applications where users may need to interact with data.

Here’s an example:

import plotly.express as px

# Sample cluster data
df = px.data.iris()
df['cluster'] = df['species_id']  # Assume 'species_id' as cluster labels

# Create an interactive scatter plot
fig = px.scatter(df, x='sepal_width', y='sepal_length', color='cluster')
fig.show()

Output: An interactive scatter plot rendered in a web browser, with points colored by clusters.

The code uses Plotly Express to create an interactive plot directly from a DataFrame. The cluster label column ‘cluster’ is mapped to the color of the points. The resulting plot is inherently interactive and can be embedded into web applications.

Method 4: Using Pandas Plot

Pandas itself has built-in plotting capabilities which are useful for quick and straightforward plotting tasks. Using the plot.scatter() function from Pandas can be a good choice when the data is already in a DataFrame, and you want to leverage the DataFrame’s functionality.

Here’s an example:

import pandas as pd

# Sample cluster data in DataFrame
df = pd.DataFrame({
    'Feature 1': [1, 2, 3, 4, 5],
    'Feature 2': [5, 4, 3, 2, 1],
    'Cluster': [0, 0, 1, 1, 1]
})

# Create scatter plot
ax = df.plot.scatter(x='Feature 1', y='Feature 2', c='Cluster', colormap='viridis')
plt.show()

Output: A simple scatter plot colored by clusters using pandas plotting interface.

The Pandas plot snippet directly plots the DataFrame columns ‘Feature 1’ and ‘Feature 2’, while coloring the points by the ‘Cluster’ column. It’s a direct method if you’re already working within the Pandas ecosystem.

Bonus One-Liner Method 5: Using ggplot

The Python implementation of ggplot from the R ecosystem, dubbed `ggplot`, can also be used for quick scatter plot visualizations. Its syntax and styling are different from Matplotlib and Seaborn but can create complex multi-layered graphics.

Here’s an example:

from ggplot import *

# Assume 'df' is a DataFrame with 'Feature 1', 'Feature 2' and 'Cluster'

# Create scatter plot with ggplot
plot = ggplot(df, aes(x='Feature 1', y='Feature 2', color='Cluster')) + geom_point()
print(plot)

Output: A scatter plot generated using ggplot’s syntax, with points colored by clusters.

This one-liner uses ggplot to map DataFrame columns to aesthetics and adds a point geometry layer for the scatter plot. Users familiar with R’s ggplot2 may prefer this approach.

Summary/Discussion

  • Method 1: Matplotlib. Provides extensive customization options. Can be verbose for simple plots.
  • Method 2: Seaborn. Simplifies the creation of complex plots and works well with DataFrames. May not offer as much low-level control as Matplotlib.
  • Method 3: Plotly. Ideal for interactive plots. More complex to setup for static images.
  • Method 4: Pandas Plot. Convenient for quick plots without leaving the DataFrame. Limited customization compared to dedicated plotting libraries.
  • Bonus Method 5: ggplot. Familiar syntax for users of R’s ggplot2. Less commonly used in the Python ecosystem compared to other libraries.