An end-to-end business analytics pipeline built on 550,000+ rows of real Brazilian e-commerce data from Olist (2016–2018).

View Interactive Tableau Dashboard

This project covers the complete analytics workflow:

• Data ingestion pipeline (Python, PostgreSQL, SQLAlchemy)
• Data cleaning and validation (pandas)
• Advanced SQL analysis (window functions, CTEs, CLV)
• KPI dashboard development
• A/B testing (scipy, p < 0.0001, n = 96,353 orders)
• Customer segmentation (K-Means, RFM model)
• Churn prediction (Logistic Regression, ROC-AUC = 1.0)
• Interactive Tableau dashboard (5 charts)

• São Paulo drove 47% of total platform revenue (R$5.2M)
• Fast delivery (≤10 days) scored 4.39/5 vs slow delivery 3.94/5 (p < 0.0001)
• 66.6% of customers churned after first purchase
• 3 customer segments identified: Champions (4,962), Loyal Customers (50,830), At Risk (37,566)

Python · PostgreSQL · SQL · pandas · scikit-learn · scipy · matplotlib · Tableau · Git

retail-analytics-pipeline/ ├── data/ │ ├── raw/ # Original Olist CSV files │ ├── cleaned/ # Cleaned datasets │ └── processed/ # Analysis-ready exports for Tableau ├── notebooks/ # Jupyter analysis notebooks ├── sql/ # Database schema and queries ├── scripts/ # Python ingestion scripts └── dashboards/ # Saved chart images

Meet Patel | M.S. Business Analytics, DePaul University LinkedIn | GitHub