🧠 AI with Python – 🐍🛍️📊 Clustering Customers using K-Means

Customer segmentation is one of the most practical applications of machine learning. By grouping customers into clusters, businesses can better understand their audience, personalize marketing strategies, and improve customer experience.

In this post, we’ll use K-Means clustering to segment customers based on Annual Income and Spending Score.

Why K-Means for Customer Segmentation?

• Unsupervised learning → no labels needed.
• Simple yet powerful → works well for 2D and higher dimensions.
• Centroids represent the “center” of each group.
• Commonly used in retail, marketing, and recommendation systems.

Loading Customer Data

For demonstration, we’ll create a small synthetic dataset with two features:

import pandas as pd

data = {
    "CustomerID": range(1, 21),
    "AnnualIncome_k$": [15, 16, 17, 18, 20, 22, 25, 28, 30, 33, 60, 62, 65, 68, 70, 72, 75, 78, 80, 85],
    "SpendingScore":   [18, 20, 22, 25, 28, 30, 35, 40, 42, 45, 55, 58, 60, 62, 65, 66, 68, 70, 72, 75],
}
df = pd.DataFrame(data)

Standardizing Features

K-Means relies on distances. To ensure features contribute fairly, we standardize them.

from sklearn.preprocessing import StandardScaler

features = ["AnnualIncome_k$", "SpendingScore"]
X = df[features].values

scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

Applying K-Means

We cluster customers into 3 groups (k=3).

from sklearn.cluster import KMeans

kmeans = KMeans(n_clusters=3, n_init=10, random_state=42)
labels = kmeans.fit_predict(X_scaled)

df["Cluster"] = labels

Visualizing Clusters

We plot Annual Income vs Spending Score, color by cluster, and mark centroids.

import matplotlib.pyplot as plt
import numpy as np

centroids_orig = scaler.inverse_transform(kmeans.cluster_centers_)

colors = np.array(["#1f77b4", "#ff7f0e", "#2ca02c"])

plt.scatter(df["AnnualIncome_k$"], df["SpendingScore"], c=colors[df["Cluster"]], s=60, edgecolor="white")
plt.scatter(centroids_orig[:, 0], centroids_orig[:, 1], c="red", s=180, marker="X", edgecolor="black", label="Centroids")
plt.xlabel("Annual Income (k$)")
plt.ylabel("Spending Score (1–100)")
plt.title("Customer Segments via K-Means (k=3)")
plt.legend()
plt.show()

Sample Output

You’ll see a scatter plot where customers are grouped into 3 clusters, each with its centroid marked by a red “X”.

Figure: Customer Segments via K-Means.

Key Takeaways

• K-Means is a go-to algorithm for customer segmentation.
• Always scale features before clustering.
• Choosing the right k is crucial → use the Elbow Method or Silhouette Score.
• Works as a foundation before moving to advanced clustering (e.g., DBSCAN, hierarchical).