An end-to-end data science project analyzing retail transaction data, segmenting customer cohorts using RFM analysis and K-Means clustering, and predicting high-value "Whale" customers to optimize targeted marketing spend and drive customer lifetime value (CLV).
- Data Processing & Integrity: Cleaned 500,000+ transaction records and filtered return credits (
InvoiceNostarting with 'C') to ensure net-positive revenue integrity. - Behavioral Phenotyping: Categorized the customer base into 3 distinct behavioral personas (Champions/Whales, Active Mid-Tier, and At-Risk) using RFM (Recency, Frequency, Monetary) metrics and K-Means Clustering.
- Actionable Marketing Strategy: Translated raw behavioral patterns into tailored strategy—from high-touch concierge retention for VIPs to surgical win-back campaigns for lapsed users.
- Predictive Analytics: Built a leak-free classification pipeline using Logistic Regression, prioritizing 72% Recall to reliably identify high-value ($1,000+ spend) customers before churn.
| Persona | Behavioral Profile | Recency (Avg) | Spend (Avg) | Marketing & Insights Strategy |
|---|---|---|---|---|
| Cluster 2 | Champions / "Whales" | ~6 Days | $85,904 | VIP Concierge Retention & Exclusive Perks |
| Cluster 0 | Active Mid-Tier | ~41 Days | $1,855 | Upsell & cross-sell to cross $1,000 spend |
| Cluster 1 | At-Risk / Lapsed | ~247 Days | $631 | Win-back campaigns with target incentives |
- 📄 Executive PDF Report:
Market_Research_Analysis_Report.pdf - 📓 Jupyter Notebook Code:
Retail_Intelligence_RFM.ipynb
Anthony A. Inzinna
M.S. Data Scientist | Behavioral Intelligence
- Email: Anthonyinzinna3@gmail.com