PythonPandasSQLiteSQLStreamlitNumPy
E-Commerce Sales & Revenue Analytics
Analyzed 120K+ rows of Brazilian e-commerce data (Olist dataset) across multiple CSVs, migrated to SQLite for relational querying, and built an interactive Streamlit dashboard to surface actionable business insights.
E
The Challenge
Raw data was spread across 8 separate CSVs with no relational structure, making cross-table analysis slow and error-prone with pure Pandas.
The Solution
Migrated all CSVs into a normalized SQLite database, enabling efficient multi-table SQL joins and aggregations that would have been impractical in-memory.
Key Outcomes
- Found repeat buyers (3% of customers) generate 2× cumulative revenue — signaling a retention gap
- Identified repeat buyers spend 9% less per order, ruling out purchase-value as the growth lever
- Mapped revenue across 27 states; São Paulo led at $5.7M total revenue
- Surfaced Northeastern states yield 85% higher revenue-per-customer ($273 vs $147)
- Built Streamlit dashboard with 5+ interactive views — segmentation, delivery performance, order distribution, and geo-revenue concentration