k Medoids Clustering in Python

In this tutorial, we will learn how to perform K medoids clustering in Python.

First, let’s import the required libraries.

import numpy as np
from scipy.spatial.distance import cdist
import matplotlib.pyplot as plt
from sklearn.datasets import make_blobs

Now we will write a function for K Medoids clustering. initially, we will select k random points as the medoids from given data points and associate each data point to the medoids by using the Euclidean distance metric then calculate the cost of which point has a minimum cost we will update it as a new medoid. At last, when there is no change in updation select those points as final medoids

def kmedoids(X, k, max_iterations=100):
    n_samples = X.shape[0]
    medoid_indices = np.random.choice(n_samples, k, replace=False)
    medoids = X[medoid_indices]

    for _ in range(max_iterations):
        # Compute distances between all points and medoids
        distances = cdist(X, medoids, metric='euclidean')
        # Assign each point to the nearest medoid
        labels = np.argmin(distances, axis=1)
        
        new_medoids = np.copy(medoids)
        
        for i in range(k):
            cluster_points = X[labels == i]
            if len(cluster_points) == 0:
                continue
            # Compute the cost for each point in the cluster to be a medoid
            cost = cdist(cluster_points, cluster_points, metric='euclidean').sum(axis=1)
            # Choose the point with the minimum cost as the new medoid
            new_medoids[i] = cluster_points[np.argmin(cost)]
        
        # if there is no change in medoids then break and return
        if np.all(new_medoids == medoids):
            break
        
        medoids = new_medoids

    return labels, medoids

Now, we will take some data points

X, _ = make_blobs(n_samples= 1000, centers=5, cluster_std=0.60, random_state=0)

Now, we will perform clustering by calling the function

k = 5
labels, medoids = kmedoids(X, k)

plot the result and cluster.

plt.figure(figsize=(6, 4))
scatter = plt.scatter(X[:, 0], X[:, 1], c=labels, cmap='viridis')
plt.scatter(medoids[:, 0], medoids[:, 1], c='red', s=50, marker='*', label='Medoids')
plt.title('K-Medoids Clustering')
plt.xlabel('X')
plt.ylabel('Y')
plt.legend()
plt.colorbar(scatter, label='Cluster')
plt.show()