Custom K-Means Clustering

This repository contains an implementation of the K-Means clustering algorithm from scratch, designed to work similarly to sklearn.cluster.KMeans. However, it uses a custom distance metric based on the dimensionality of the input data:

• 1D Data → Uses Manhattan Distance (L1 norm)
• 2D Data → Uses Euclidean Distance (L2 norm)
• 3D and above → Cosine Distance

Example Output

when k=3

Features

✅ Fully implemented from scratch ✅ Supports multiple distance metrics based on input dimensionality ✅ Works like Scikit-learn’s KMeans

Installation

You can install this package by cloning the repository and using pip:

pip install .

Or you can install it from pypi

pip install simple-kmeans

Usage

Here’s an example of how to use the custom K-Means implementation:

from simple_kmeans import KMeans
import numpy as np
import matplotlib.pyplot as plt

# Generate sample data
X = np.array([[1, 2], [1.5, 1.8], [5, 8], [8, 8], [1, 0.6], [9, 11]])

# Initialize and fit the model
kmeans = KMeansCustom(k=2, max_iters=100)
kmeans.fit(X)

# Get cluster centers and labels
print("Cluster Centers:", kmeans.centroids)
print("Labels:", kmeans.labels)

# Plot results
plt.scatter(X[:, 0], X[:, 1], c=kmeans.labels, cmap='viridis')
plt.scatter(kmeans.centroids[:, 0], kmeans.centroids[:, 1], marker='x', c='red', s=200, label='Centroids')
plt.legend()
plt.title("Custom K-Means Clustering")
plt.show()

License

This project is open-source and available under the MIT License.

Contributions

Contributions, issues, and feature requests are welcome! Feel free to fork this repository and submit a pull request.