This project demonstrates a full-cycle, professional data analytics workflow, transforming raw transactional data into a high-impact, actionable strategic plan. The analysis uses a hybrid machine learning model to segment customers and concludes with an interactive Power BI dashboard to present the findings.
Live Dashboard: [https://bitmesra-my.sharepoint.com/:u:/g/personal/btech10456_22_bitmesra_ac_in/Ea2i6s8VyhpAhvVidXyQVlwBwJPcu9oLij_292iewOMOXA?e=6XzQk3]
Dashboard Preview:
This project utilizes a three-tiered architecture to mirror a professional data environment:
- Data Warehouse & Transformation: SQL (via SQLite) for data storage and calculation of RFM metrics using advanced CTEs.
- Advanced Analytics & Machine Learning: Python (Pandas, Scikit-learn) for implementing a hybrid DBSCAN + K-Means clustering model and validating the results with a Silhouette Score.
- Business Intelligence & Storytelling: Microsoft Power BI for creating a fully interactive dashboard that visualizes key insights and strategic recommendations.
The analysis successfully identified four distinct and statistically significant customer personas:
| Persona | Data Profile |
|---|---|
| High-Value Reseller | Very recent, highly frequent, and high-spending core customers. |
| Loyal Homemaker | The largest group of consistent, moderately recent and frequent customers. |
| At-Risk Shopper | A large group of customers who have not purchased in a long time and are about to churn. |
| Anomalous Shopper | A small, high-value outlier group with erratic buying patterns. |
Based on the personas, a multi-faceted strategy was developed to enhance customer engagement and maximize revenue:
| Persona | Recommended Action |
|---|---|
| High-Value Reseller | Launch a "Business Program" with bulk pricing and exclusive product previews. |
| Loyal Homemaker | Introduce personalized "Smart Bundles" and a "Subscribe & Save" feature. |
| At-Risk Shopper | Execute a targeted "Win-Back" campaign with personalized, limited-time offers. |
| Anomalous Shopper | Isolate and Investigate for potential B2B opportunities or fraudulent activity. |
- The Jupyter Notebook
customer_segmentation.ipynbcontains all the Python code for the analysis. - The
rfm_query.sqlfile contains the advanced SQL query used for data transformation. - The final
customer_segments_final.csvdataset is provided to allow for easy reproduction of the Power BI dashboard.