Online Retail Analysis & Customer Segmentation

A comprehensive data analysis project examining e-commerce transactions to uncover revenue trends, customer segments, and actionable business insights using machine learning and statistical methods.

Overview

This project analyzes the Online Retail II dataset from the UCI Machine Learning Repository, containing over 1 million transactions from a UK-based online retailer (2009-2011). Through RFM analysis and K-means clustering, we segment customers into actionable personas and identify strategic opportunities for business growth.

Key Deliverables:

• Customer segmentation using RFM analysis (4 distinct clusters)
• Revenue trend analysis and seasonality patterns
• Geographic market distribution insights
• Product performance analytics
• 90-day revenue forecasting using ARIMA

Key Findings

Business Metrics

• Total Revenue: £17,743,429
• Customer Base: 5,878 unique customers
• Transaction Volume: 1,067,371 records
• Geographic Reach: 38 countries

Customer Segmentation Results

Segment	Count	% of Base	Avg Spend	Total Revenue	Revenue %
Ultra VIP	4	0.1%	£437K	£1.7M	9.9%
High-Value	35	0.6%	£83K	£2.9M	16.4%
Active Regular	3,841	65.3%	£3K	£11.6M	65.4%
At-Risk	1,998	34.0%	£765	£1.5M	8.5%

Critical Insight: 39 customers (0.7% of base) generate £4.6M (26.3% of total revenue)

Market Concentration

• UK: 83.0% of revenue (£14.7M) - extreme geographic concentration
• Australia: 1.0% (£170K) - largest non-UK market
• International markets: Only 17% combined, indicating expansion opportunity

Product Analysis

• Top 10 products: 8.5% of revenue (healthy diversification)
• Leading product: REGENCY CAKESTAND 3 TIER (£286K)
• Category dominance: Home décor and gift items

Visualizations

Monthly Revenue Trend

Customer Segmentation

Top Products

Clustering Validation

Technology Stack

• Python 3.10+
• pandas, numpy - Data manipulation
• scikit-learn - Machine learning (K-Means clustering)
• matplotlib, seaborn - Visualizations
• plotly - Interactive maps
• pmdarima - Time series forecasting
• Jupyter Notebook - Interactive analysis

Getting Started

Prerequisites

• Python 3.10 or higher
• pip package manager
• Kaggle account (for dataset access)

Installation

1. Clone the repository

   git clone https://github.com/itsaryanchauhan/online-retail-analysis.git
   cd online-retail-analysis

2. Install dependencies

   pip install -r requirements.txt

3. Configure Kaggle credentials

   mkdir ~/.kaggle
   mv kaggle.json ~/.kaggle/
   chmod 600 ~/.kaggle/kaggle.json

Reproducibility

This analysis is designed for reproducible results:

• K-Means clustering uses random_state=42 for consistent cluster assignments
• Data processing pipeline is deterministic (no random sampling)
• Same dataset and preprocessing steps will yield identical customer segments
• ARIMA forecasting may show minor variations due to optimization algorithms

To ensure identical results:

import numpy as np
np.random.seed(42)  # Set before running analysis

Usage

Run complete analysis pipeline:

python3 run_analysis.py

Or from the scripts directory:

cd scripts
python3 main.py

Interactive Jupyter Notebook:

jupyter notebook retail_analysis.ipynb

Use individual modules:

import sys
sys.path.append('scripts')

from data_loader import load_retail_data
from rfm_analysis import calculate_rfm, segment_customers

df = load_retail_data()
rfm = calculate_rfm(df)
rfm = segment_customers(rfm, n_clusters=4)

Project Structure

online-retail-analysis/
├── retail_analysis.ipynb   # Complete analysis notebook
├── scripts/
│   ├── data_loader.py       # Data loading utilities
│   ├── data_cleaning.py     # Preprocessing & feature engineering
│   ├── rfm_analysis.py      # Customer segmentation
│   ├── visualizations.py    # Chart generation
│   └── main.py              # Pipeline orchestrator
├── plots/                  # Generated visualizations
├── requirements.txt        # Dependencies
└── README.md              # Documentation

Business Recommendations

High Priority (0-3 months)

1. Protect high-value customers: 39 customers generate 26.3% of revenue
2. Launch Q4 marketing campaigns (September-November peak season)
3. Deploy win-back campaigns for 1,998 dormant customers

Medium Term (3-6 months)

1. Reduce UK dependency from 83% through international expansion
2. Implement tiered loyalty program for active customers
3. Expand product portfolio beyond home décor

Long Term (6-12 months)

1. Build predictive churn models
2. Develop subscription-based revenue streams
3. Systematic market entry into Australia, Germany, Netherlands

Projected Impact: £2.5M-£3.5M additional revenue (14-20% growth)

Resources

• Jupyter Notebook (Google Colab): Open in Colab
• Dataset: Online Retail II UCI
• Kaggle Notebook: View on Kaggle
• GitHub Repository: Source Code

Results & Outputs

Running the analysis generates:

Visualizations (plots/)

• monthly_revenue.png
• top_products.png
• rfm_clusters.png
• elbow_silhouette.png

Author

Aryan Chauhan
GitHub | LinkedIn

Data Analyst specializing in customer analytics and business intelligence.

Contributing

Contributions are welcome. Please fork the repository and submit a pull request with your improvements.

Potential enhancements:

• Cohort analysis and retention metrics
• Advanced forecasting models (Prophet, LSTM)
• Interactive dashboard (Streamlit/Dash)
• A/B testing framework

License

This project is available under the MIT License.

Acknowledgments

• UCI Machine Learning Repository for the dataset
• Kaggle community for data science resources
• scikit-learn, pandas, and visualization library contributors