Customer Segmentation Pipeline with RFM, K-Means & BI Dashboards

Project demonstrates a full analytics pipeline for eCommerce customer segmentation using RFM analysis and K-Means clustering, followed by interactive Power BI dashboards.

Project Overview

The goal is to analyze customer behavior and identify meaningful segments to support business decisions. The process includes:

• Cleaning and transforming raw transactional data
• Building RFM metrics (Recency, Frequency, Monetary)
• Performing customer segmentation via clustering (K-Means)
• Visualizing insights through interactive dashboards

End-to-End Data Pipeline

This project represents a complete beginner-friendly data pipeline:

1. Data Ingestion → from raw Excel file (OnlineRetail.xlsx)
2. Data Cleaning → handled with Python (nulls, outliers, types)
3. Feature Engineering → RFM metrics computed per CustomerID
4. Machine Learning → KMeans clustering to segment customers
5. Storage → SQLite used to store cleaned datasets
6. Visualization → Power BI (and optionally Tableau) used for reporting

Tech Stack

• Python: pandas, numpy, scikit-learn, matplotlib • Power BI • Tableau • SQLite

Dashboards

Power BI Dashboard

RFM & Segmentation Overview: • KPIs: Total Revenue, Customers, Orders • RFM Heatmap • RFM Segment Distribution • Country Map • Top Customers Customer Behavior: • Customer distribution by country • Invoice time of day analysis • Hourly heatmap / customer activity • Customer type breakdown Segment Deep Dive: • Segment vs Country matrix • Radar chart: avg. RFM metrics per segment • Line & stacked chart: Active customer behavior across segments

Tableau Dashboard

• Segment Distribution: Interactive bar chart of customer count by RFM segment or score. • RFM Heatmap: Grid showing customer density by Recency and Frequency quartiles. • Country Map: Bubble map of total revenue by country. • RFM Segment Chart: Highlights customer distribution across RFM segments. • ML Target by Country: Shows predicted target rates by country. • Customer Flow: Sankey-style chart of segment transitions by ML target prediction.

What I Learned

• Cleaning and transforming real-world transactional data • Designing customer value metrics • Using machine learning in business context • Building visual dashboards for decision-making • Structuring end-to-end data projects • Business Intelligence

Author: Jakub
Educational portfolio project