👥 Customer Segmentation — RFM Analysis

🔎 Overview

This project performs RFM analysis (Recency, Frequency, Monetary) to segment customers based on their purchasing behavior.
Goal: identify Loyal, New, At Risk, and Lost customers and quantify their value to the business.

Dataset:

Online Retail II

---

🧰 Tech Stack

• Python: pandas, numpy
• Visualization: matplotlib, seaborn
• Environment: Jupyter Notebook

---

📁 Project Structure

Customer_Segmentation_RFM/ │── data/ │ └── online_retail.xlsx │── images/ │ └── rfm_segments.png │── notebooks/ │ └── rfm_analysis.ipynb │── requirements.txt └── README.md

---

🧪 Methodology

1. Clean data: remove returns/duplicates, filter positive quantities, compute TotalPrice.
2. Build RFM table at the CustomerID level:
   • R (Recency): days since last purchase
   • F (Frequency): number of invoices
   • M (Monetary): total spend
3. Score each metric into quartiles → R_Score, F_Score, M_Score.
4. Combine into RFM_Score and map to Segments (Loyal, New, At Risk, Lost, Regular).
5. Visualize distribution of segments and summarize average R/F/M per segment.

---

📊 Key Insights (replace with your results)

• Loyal customers show the highest Monetary value and above-average Frequency.
• At Risk customers have long Recency and still meaningful Monetary — targeted reactivation can pay off.
• A small VIP-like group (high F & M) contributes disproportionally to revenue (Pareto pattern).
• New customers can be guided to the next purchase via cross-sell bundles.

---

🖼 Visualizations

---

🚀 How to Run

git clone https://github.com/Blladerunner/customer-segmentation-rfm-analysis.git
cd customer-segmentation-rfm-analysis

python -m venv .venv
.\.venv\Scripts\Activate.ps1    # Windows
# source .venv/bin/activate     # macOS/Linux

pip install -r requirements.txt
python -m notebook notebooks/rfm_analysis.ipynb

🔙 Back to Portfolio