I analyzed the Brazilian e-commerce dataset (O-list) using SQL (Duck DB in Jupyter Notebook) and Power BI to uncover valuable business insights.
The comprehensive dataset includes 100k+ orders, 3k sellers, and 70k products from 2016–2018, with data on customers, orders, payments, reviews, sellers, products, and geolocation.
Data & Tools: Olist dataset (orders, customers, reviews), SQL, Power BI
Key Analyses:
- Revenue concentration in specific categories (Health/Beauty, Watches, Home Goods).
- High freight costs (~15% of order value) presenting a logistics optimization opportunity.
- Clear sales growth throughout 2017–2018 with seasonal patterns (holiday and year-end peaks).
- Geographic insights showing sales concentrated in São Paulo & southeast Brazil, while sellers have broader distribution.
- Payment behavior: Customers frequently use 2–12 installment plans, affecting cash flow timing.
- Review analysis: Approximately 60% of orders receive 5-star ratings, with delivery delays strongly linked to lower ratings.
Deliverables: