Customer Segmentation Clusters Using K-Means and PCA 

In this project, I acted as a consultant for a bank in New York City to help the marketing team launch a highly targeted advertising campaign. The goal was to use customer data collected over six months to segment the customer base into at least three actionable groups. This segmentation would enable the bank to personalize marketing strategies, driving engagement and increasing revenue. 

Data Preparation:

• Cleaned and prepared the dataset by handling missing values and normalizing numerical features.

• Key data points included balances, spending behaviors, credit limits, and payment patterns.

Clustering Analysis:

• Applied K-Means Clustering, an unsupervised learning algorithm, to group customers with similar behaviors.

• Determined the optimal number of clusters using the Elbow Method, ensuring precise segmentation.

Dimensionality Reduction:

• Utilized Principal Component Analysis (PCA) to simplify the data while retaining meaningful insights, improving the clarity and effectiveness of clustering.

Visualization:

• Created easy-to-interpret charts and graphs to present the segmentation results, showcasing distinct customer groups.

Key Outputs

• Identified three distinct customer segments:

  • High Spenders: Customers with significant credit usage and one-off purchases.

  • Frequent Purchasers: Customers with consistent transaction activity, often using installment plans.

  • Low Engagement Customers: Customers with low balances and minimal account activity.

• Delivered actionable insights to guide marketing strategies tailored to each group.

• Provided a data-driven framework for designing personalized marketing campaigns.

Outcome

This project demonstrates my ability to:

• Use data science techniques like clustering and dimensionality reduction to solve real-world business problems.

• Provide actionable insights through clear, effective communication and visualization.

• Support businesses in achieving strategic goals by leveraging data-driven decision-making.

Click here to visit the GitHub repository for more details: GitHub Repository