1. Executive Summary

In this project, I analyzed a transactional database of nearly 400,000 records for a UK-based online retail company. The primary objective was to transition from raw transactional data to actionable business intelligence. Using Python for Data Engineering, I cleaned anomalies, handled edge cases, and engineered an RFM (Recency, Frequency, Monetary) model to segment the customer base.

The analysis revealed critical business insights: 15.3% of the customer base (VIPs) drives over 75% of total revenue, indicating a strong Pareto distribution. Furthermore, a deep dive into Average Order Value (AOV) uncovered a hidden B2B wholesale segment operating within the B2C platform. I provided data-driven recommendations to marketing stakeholders to mitigate churn and maximize the Lifetime Value (LTV) of key accounts.

2. Tech Stack & Tools

• Language: Python 3
• Data Manipulation: Pandas, NumPy
• Data Visualization: Matplotlib
• Methodologies: Data Cleaning, Feature Engineering, RFM Modeling, Cohort Analysis.

3. Methodology & Data Engineering

Before modeling, rigorous data cleaning was performed to ensure financial accuracy:

• Anomaly Detection: Identified and removed a "fake whale" transaction (a massive cancellation masquerading as a purchase) that would have skewed the Monetary metrics.
• Financial Integrity: Filtered out transactions with UnitPrice = 0.0. Upon investigation, these were identified as free samples or replacements sent exclusively to 'Champion' clients, not actual revenue-generating sales.
• Feature Engineering: Created the Total_Dinero feature and transformed individual line-item transactions into an aggregated customer-level RFM matrix.
• Scoring System: Applied quintile-based scoring (pd.qcut) to assign a 1-5 rank across Recency, Frequency, and Monetary values, categorizing clients into 5 actionable business segments.

4. Key Business Insights & Recommendations

🥇 Insight 1: The Pareto Principle in Action (VIP Impact)

• The Data: VIP clients ('Champions') make up only 15.34% (75 clients) of the customer base, but they drive a massive 75.1% ($1,211,783) of total historical sales.
• Business Recommendation: Implement a dedicated Key Account Management (KAM) program for these 75 clients. The financial concentration risk is extremely high; losing even a fraction of this segment would severely impact the company's bottom line.

<aside> 💡

</aside>