This project demonstrates unsupervised clustering techniques applied to a UK bank's customer dataset. The primary aim is to segment customers into distinct groups based on demographic and financial behavior, enabling data-driven marketing and service personalization.
We applied Hierarchical Clustering and Two-Step Clustering using SPSS to identify natural customer segments. Key objectives:
- Discover distinct customer profiles.
- Understand variable contributions to cluster formation.
- Use clustering results to recommend marketing strategies.
Source: Simulated UK bank customer dataset (425 records)
Variables Used for Clustering (standardized via Z-scores):
Age
Current Account Balance
Savings Account Balance
Months as Customer
Months Employed
- Method: Ward’s Method
- Distance Metric: Squared Euclidean
- Output: Dendrogram
📌 Result:
- 4 clusters identified
- Visual splits observed in dendrogram at sharp linkage jumps
🔢 Proximity Matrix (Top View)
(Excerpt from squared Euclidean distance between select cases)
- Distance: Euclidean
- Auto-clustering: Enabled
- Output: Cluster Quality, Model Summary
📈 Cluster Quality:
- Silhouette Score: ~0.4 (Fair)
📊 Model Summary:
Cluster | Description |
---|---|
1 | Younger customers, low balances, short job tenure |
2 | Mature, high-income customers with long tenure |
3 | Mid-aged customers with moderate finances |
4 | High current balance, but low savings and tenure |
📋 Case Processing Summary
All 425 records included and processed:
- Cluster 2: Target with premium and wealth management services
- Cluster 1: Offer beginner-friendly digital banking tools
- Cluster 3: Promote long-term savings and credit plans
- Cluster 4: Improve retention through loyalty and engagement offers
data.xlsx
– Raw customer datacluster.docx
– Detailed reportimages/
– SPSS chart exports
This analysis highlights the value of unsupervised machine learning in marketing strategy. By combining Hierarchical Clustering for visual insights and Two-Step Clustering for profiling strength, we provide actionable segmentation that supports tailored customer engagement.
🚀 Built with SPSS, Excel, and GitHub to showcase real-world data segmentation in financial services.