Practical Analysis of Market Customer Segmentation Based on K-Means Clustering

This project is a complete practical case of market customer segmentation analysis. It uses the K-Means clustering algorithm combined with PCA dimensionality reduction technology to segment customers using the Mall Customers dataset, supporting the formulation of precision marketing strategies. The project is from GitHub author Durgaprasad995852, released on June 9, 2026, and features localization adaptation for the Indian market, a complete machine learning pipeline, and clear business application value. The following floors will introduce the project background, methods, results, applications, and other content in detail.

In a highly competitive business environment, traditional one-size-fits-all marketing can hardly meet personalized needs. Customer segmentation helps enterprises formulate differentiated strategies by dividing groups with different characteristics, improving customer satisfaction and revenue. This project demonstrates a complete customer segmentation process, using K-Means to analyze the Mall Customers dataset and combining PCA visualization to support precision marketing decisions.

Technology Stack: Uses Python ecosystem tools including Pandas (data processing), NumPy (numerical computation), Scikit-learn (K-Means/PCA models), Matplotlib/Seaborn (visualization), and Joblib (model persistence). Project Structure: Includes directories such as data (raw data), models (model storage), outputs (result outputs), and src (core code modules). Core Methods:

• Uses RFM analysis thinking to evaluate customer value;
• K-Means algorithm: Initialization → Assignment → Update → Iterative convergence;
• Elbow method to determine the optimal number of clusters;
• PCA dimensionality reduction: Compresses features to 2-3 dimensions for easy visualization while retaining key variance information.

Indian Localization Adaptation: For the Indian Rupee (INR) format of annual income in the original data (e.g., ₹1,20,000), convert the format to numerical type while retaining the formatted version in reports, and standardize features to eliminate dimensionality effects. Feature Engineering: Uses key features such as customer age, gender, annual income (after numerical conversion), and spending score.

Visualization Outputs: Generates elbow method plots (to determine optimal K value), PCA scatter plots (to intuitively show cluster distribution), and cluster silhouette plots (to show cluster boundaries and density). Model Output Files: Includes customer data with cluster labels (clustered_customers.csv), cluster statistical report (cluster_report.csv), elbow method plot (elbow_method.png), PCA coordinate data (pca_coordinates.csv), PCA cluster visualization plot (pca_clusters.png), and persisted model (kmeans_model.pkl).

Precision Marketing: Offer VIP discounts to high-spending customers, targeted ads to potential customers, and promotions to price-sensitive customers; Product Strategy: Adjust product mix, develop customized products, and optimize inventory to match demand; Customer Relationship Management: Identify customers at risk of churning, implement differentiated service strategies, and increase customer lifetime value (CLV).

Technical Highlights: Complete ML pipeline, interpretability (PCA + visualization), Indian market localization adaptation, engineering practices (good structure and dependency management), and business orientation (directly serving marketing decisions). Expansion Suggestions: Expand customer behavior features (purchase frequency, recent purchase time), compare algorithms like DBSCAN/hierarchical clustering, deploy the model as an API to support real-time segmentation, and establish a dynamic update mechanism to adapt to changes in customer behavior.